From 0c9fde4b76be0d8636a8c5521b5975ece77db5eb Mon Sep 17 00:00:00 2001 From: Benedikt Heidrich Date: Thu, 8 Aug 2024 14:56:35 +0200 Subject: [PATCH] [DOC] Add example notebook for the graphical pipeline (#5175) #### Reference Issues/PRs See #4652 and #4281 #### What does this implement/fix? Explain your changes. This PR adds the graphical pipeline jupyter notebook. Furthermore, it ensures that the ForecastingGridSearchCV call fit,predict etc. using keywords instead of positional arguments to enable that the gridsearch works with the graphical pipeline. --- examples/05_graphical_pipelines.ipynb | 5139 +++++++++++++++++ examples/img/classification_pipeline.png | Bin 0 -> 22850 bytes examples/img/forecasting_pipeline.png | Bin 0 -> 24834 bytes examples/img/graphical_pipeline.png | Bin 0 -> 21525 bytes examples/img/graphical_pipeline_example.png | Bin 0 -> 75478 bytes .../img/graphical_pipeline_example_grid.png | Bin 0 -> 82335 bytes .../img/graphical_pipeline_simplified.png | Bin 0 -> 80681 bytes examples/img/sequential_pipeline.png | Bin 0 -> 8359 bytes .../model_evaluation/_functions.py | 2 +- 9 files changed, 5140 insertions(+), 1 deletion(-) create mode 100644 examples/05_graphical_pipelines.ipynb create mode 100644 examples/img/classification_pipeline.png create mode 100644 examples/img/forecasting_pipeline.png create mode 100644 examples/img/graphical_pipeline.png create mode 100644 examples/img/graphical_pipeline_example.png create mode 100644 examples/img/graphical_pipeline_example_grid.png create mode 100644 examples/img/graphical_pipeline_simplified.png create mode 100644 examples/img/sequential_pipeline.png diff --git a/examples/05_graphical_pipelines.ipynb b/examples/05_graphical_pipelines.ipynb new file mode 100644 index 00000000000..bc7e2688cb5 --- /dev/null +++ b/examples/05_graphical_pipelines.ipynb @@ -0,0 +1,5139 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "# Graphical Pipelines\n", + "\n", + "### Content of this Notebook:\n", + "* Understanding what are graphical pipelines\n", + "* Understanding the API of graphical pipelines\n", + "* Examples of simple pipelines and how they can be implemented with graphical pipelines.\n", + "* More complex graphical pipeline (Forecasting + GridSearch)\n", + "* Grid search with a graphical pipeline\n", + "\n", + "\n", + "\n", + "### What are Graphical Pipelines?\n", + "Recap sequential pipelines:\n", + "\n", + "\n", + "\n", + "Many tasks are non-sequential. To solve this two possibilities exist:\n", + "1. Nesting Sequential Pipelines.\n", + "2. Using Graphical Pipelines.\n", + "\n", + "\n", + "Thus, there is the generalised graphial pipeline.\n", + "* Graphical means that different steps may share the same predecessor or provide their outputs to the same successor (the dataflows can branch and merge).\n", + "\n", + "\n", + "\n", + "* Generalised means that the pipeline can be used for multiple tasks (e.g. forecasting, classification, ...).\n", + "\n", + "**Note**\n", + "\n", + "The graphical pipeline is a new feature, Thus, if you are considering any issues, we would be happy to get feedback on the graphical pipeline.\n", + "\n", + "\n", + "### Potential Use-Cases\n", + "There exist various potential use-case for the graphical pipeline. In the following, we focus on a forecasting and a classification pipeline.\n", + "#### Forecasting Use-Case for Graphical Pipelines\n", + "\n", + "\n", + "The input of forecasters depends on the output of other forecasters, which same the same input.\n", + "* Forecaster could use the same preprocessing (branching of data flow)\n", + "* Forecaster could use outputs of multiple predeccessors (merging of data flow)\n", + "\n", + "\n", + "\n", + "\n", + "**Note:** The current experimental state of the graphical pipeline does not fully support this use-case. However, we are working on this. If you are interested in this use-case and want to contribute, please contact us.\n", + "\n", + "### Credits\n", + "The graphical pipeline was first developed by pyWATTS [1] and was then adapted for sktime. The original implementation can be found [pyWATTS](https://github.com/KIT-IAI/pyWATTS). pyWATTS is a open source library developed at the Institute of Applied Informatics and Automation at the KIT and funded by HelmholtzAI.\n", + "\n", + "> [1] Heidrich, Benedikt, et al. \"pyWATTS: Python workflow automation tool for time series.\" arXiv preprint arXiv:2106.10157 (2021)." + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "\n", + "**Note:** The current experimental state of the graphical pipeline does not fully support this use-case. However, we are working on this. If you are interested in this use-case and want to contribute, please contact us.\n", + "\n" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "## How to build a Graphical Pipeline\n", + "\n", + "Let us first visualise a simple forecasting pipeline, we want to construct: \n", + "\n", + "\n", + "\n", + "\n", + "\n", + "Then we are having to ways on how to construct this pipeline with the graphical pipeline\n", + "\n", + "1. Pass all steps to the pipeline during initialisation as for the sequential pipelines.\n", + "\n" + ] + }, + { + "cell_type": "code", + "execution_count": 1, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n" + ] + } + ], + "source": [ + "from sktime.forecasting.sarimax import SARIMAX\n", + "from sktime.pipeline.pipeline import Pipeline\n", + "from sktime.transformations.series.difference import Differencer\n", + "\n", + "differencer = Differencer()\n", + "\n", + "general_pipeline = Pipeline(\n", + " [\n", + " {\"skobject\": differencer, \"name\": \"differencer\", \"edges\": {\"X\": \"y\"}},\n", + " {\n", + " \"skobject\": SARIMAX(),\n", + " \"name\": \"sarimax\",\n", + " \"edges\": {\"X\": \"X\", \"y\": \"differencer\"},\n", + " },\n", + " {\n", + " \"skobject\": differencer,\n", + " \"name\": \"differencer_inv\",\n", + " \"edges\": {\"X\": \"sarimax\"},\n", + " \"method\": \"inverse_transform\",\n", + " },\n", + " ]\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "2. Create a pipeline object and add the steps one by one.\n", + "\n", + "\n" + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n" + ] + } + ], + "source": [ + "\n", + "general_pipeline = Pipeline()\n", + "differencer = Differencer()\n", + "\n", + "general_pipeline = general_pipeline.add_step(\n", + " differencer, \"differencer\", edges={\"X\": \"y\"}\n", + ")\n", + "general_pipeline = general_pipeline.add_step(\n", + " SARIMAX(), \"sarimax\", edges={\"X\": \"X\", \"y\": \"differencer\"}\n", + ")\n", + "general_pipeline = general_pipeline.add_step(\n", + " differencer, \"differencer_inv\", edges={\"X\": \"sarimax\"}, method=\"inverse_transform\"\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\n", + "## Explanation of the parameters\n", + "\n", + "The `add_step`'s parameter or key of the dicts in the step list during initialisation are:\n", + "\n", + "* skobject: The sktime object added to the pipeline\n", + "* name: The name of the step\n", + "* edges: The keys of the dictionary indicate the input of the skobject (X or y), and the values are the names of the steps that should be connected to the input argument. Note subsetting using `__` and feature union via lists are supported.\n", + "* method: The skobject's method that should be called. If not provided, the default method would be inferred based on the added skobject. This parameter is used for the inverse_transform method. Optional.\n", + "* kwargs: Additional keyword arguments passed to the sktime object. Optional.\n" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Now let us fit the pipeline and make a prediction" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": {}, + "outputs": [ + { + "data": { + "text/plain": [ + "1959 67213.735362\n", + "1960 68328.076310\n", + "1961 68737.861398\n", + "1962 71322.894026\n", + "Freq: A-DEC, Name: TOTEMP, dtype: float64" + ] + }, + "execution_count": 3, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "from sktime.datasets import load_longley\n", + "from sktime.forecasting.model_selection import temporal_train_test_split\n", + "\n", + "y, X = load_longley()\n", + "y_train, y_test, X_train, X_test = temporal_train_test_split(y, X)\n", + "\n", + "general_pipeline.fit(y=y_train, X=X_train, fh=[1, 2, 3, 4])\n", + "general_pipeline.predict(X=X_test)" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "## Further Examples\n" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "### Classification Pipeline\n", + "A simple classification pipeline implemented using the graphical pipeline." + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n" + ] + } + ], + "source": [ + "from sktime.classification.distance_based import KNeighborsTimeSeriesClassifier\n", + "from sktime.transformations.series.exponent import ExponentTransformer\n", + "\n", + "general_pipeline = Pipeline()\n", + "general_pipeline = general_pipeline.add_step(\n", + " ExponentTransformer(), \"exponent\", edges={\"X\": \"X\"}\n", + ")\n", + "general_pipeline = general_pipeline.add_step(\n", + " KNeighborsTimeSeriesClassifier(), \"classifier\", edges={\"X\": \"exponent\", \"y\": \"y\"}\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Or alternatively defined using the constructor API." + ] + }, + { + "cell_type": "code", + "execution_count": 5, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "general_pipeline = Pipeline(\n", + " [\n", + " {\"skobject\": ExponentTransformer(), \"name\": \"exponent\", \"edges\": {\"X\": \"X\"}},\n", + " {\n", + " \"skobject\": KNeighborsTimeSeriesClassifier(),\n", + " \"name\": \"classifier\",\n", + " \"edges\": {\"X\": \"exponent\", \"y\": \"y\"},\n", + " },\n", + " ]\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "This pipeline can be visualised as follows:\n", + "\n", + "\n" + ] + }, + { + "cell_type": "code", + "execution_count": 6, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "data": { + "text/plain": [ + "array(['0', '1', '2', '0', '1', '2', '0', '1', '2', '0', '1', '2', '0',\n", + " '1', '2', '0', '1', '2', '0', '1', '2', '0', '1', '2', '0', '1',\n", + " '2', '0', '1', '2', '0', '1', '2', '0', '1', '2'], dtype='\n", + "\n", + "\n", + "The data is taken from the macrodata dataset from the statsmodels package.\n" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "**Note** We stick with the add_step in the following.\n", + "\n", + "Create Graphical Pipeline Instance" + ] + }, + { + "cell_type": "code", + "execution_count": 7, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n" + ] + }, + { + "data": { + "text/html": [ + "
Pipeline()
Please rerun this cell to show the HTML repr or trust the notebook.
" + ], + "text/plain": [ + "Pipeline()" + ] + }, + "execution_count": 7, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "pipe = Pipeline()\n", + "pipe.set_config(warnings=\"off\")\n" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Add Preprocessing\n" + ] + }, + { + "cell_type": "code", + "execution_count": 8, + "metadata": {}, + "outputs": [], + "source": [ + "from sklearn.preprocessing import StandardScaler\n", + "\n", + "from sktime.transformations.series.adapt import TabularToSeriesAdaptor\n", + "from sktime.transformations.series.detrend import Deseasonalizer\n", + "\n", + "pipe = pipe.add_step(\n", + " TabularToSeriesAdaptor(StandardScaler()),\n", + " name=\"scaler\",\n", + " edges={\"X\": \"X__realgdp_realdpi_unemp\"},\n", + ")\n", + "pipe = pipe.add_step(\n", + " Deseasonalizer(sp=4), name=\"deseasonalizer\", edges={\"X\": \"X__realgdp_realdpi\"}\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Add forecastesr for GDP and DPI" + ] + }, + { + "cell_type": "code", + "execution_count": 9, + "metadata": {}, + "outputs": [], + "source": [ + "from sklearn.linear_model import Lasso, Ridge\n", + "\n", + "from sktime.forecasting.compose import make_reduction\n", + "\n", + "pipe = pipe.add_step(\n", + " make_reduction(Ridge(), windows_identical=False, window_length=5),\n", + " name=\"forecaster_gdp\",\n", + " edges={\"y\": \"deseasonalizer__realgdp\"},\n", + ")\n", + "\n", + "pipe = pipe.add_step(\n", + " make_reduction(Ridge(), windows_identical=False, window_length=5),\n", + " name=\"forecaster_dpi\",\n", + " edges={\"y\": \"deseasonalizer__realdpi\"},\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Add Forecaster for unemployment rate that depends on forecasts of GDP and DPI" + ] + }, + { + "cell_type": "code", + "execution_count": 10, + "metadata": {}, + "outputs": [], + "source": [ + "pipe = pipe.add_step(\n", + " make_reduction(Ridge(), windows_identical=False, window_length=5),\n", + " name=\"forecaster_unemp\",\n", + " edges={\n", + " \"y\": \"scaler__unemp\",\n", + " \"X\": [\n", + " \"forecaster_gdp\",\n", + " \"forecaster_dpi\",\n", + " ],\n", + " },\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Add forecaster for the inflation that depends on forecasted DPI and unemployment rate\n", + "\n" + ] + }, + { + "cell_type": "code", + "execution_count": 11, + "metadata": {}, + "outputs": [], + "source": [ + "pipe = pipe.add_step(\n", + " make_reduction(Ridge(), windows_identical=False, window_length=5),\n", + " name=\"forecaster_inflation\",\n", + " edges={\"X\": [\"forecaster_dpi\", \"forecaster_unemp\"], \"y\": \"y\"},\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Load data and split them into train and test" + ] + }, + { + "cell_type": "code", + "execution_count": 12, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
realgdprealdpiunemp
Period
1959Q12710.3491886.95.8
1959Q22778.8011919.75.1
1959Q32775.4881916.45.3
1959Q42785.2041931.35.6
1960Q12847.6991955.55.2
............
2005Q312683.1539308.05.0
2005Q412748.6999358.74.9
2006Q112915.9389533.84.7
2006Q212962.4629617.34.7
2006Q312965.9169662.54.7
\n", + "

191 rows × 3 columns

\n", + "
" + ], + "text/plain": [ + " realgdp realdpi unemp\n", + "Period \n", + "1959Q1 2710.349 1886.9 5.8\n", + "1959Q2 2778.801 1919.7 5.1\n", + "1959Q3 2775.488 1916.4 5.3\n", + "1959Q4 2785.204 1931.3 5.6\n", + "1960Q1 2847.699 1955.5 5.2\n", + "... ... ... ...\n", + "2005Q3 12683.153 9308.0 5.0\n", + "2005Q4 12748.699 9358.7 4.9\n", + "2006Q1 12915.938 9533.8 4.7\n", + "2006Q2 12962.462 9617.3 4.7\n", + "2006Q3 12965.916 9662.5 4.7\n", + "\n", + "[191 rows x 3 columns]" + ] + }, + "execution_count": 12, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "from sktime.datasets import load_macroeconomic\n", + "from sktime.forecasting.base import ForecastingHorizon\n", + "\n", + "data = load_macroeconomic()\n", + "\n", + "X = data[[\"realgdp\", \"realdpi\", \"unemp\"]]\n", + "y = data[[\"infl\"]]\n", + "fh = ForecastingHorizon([1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12])\n", + "\n", + "y_train, y_test, X_train, X_test = temporal_train_test_split(y, X=X, fh=fh)\n", + "X_train" + ] + }, + { + "cell_type": "code", + "execution_count": 13, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
infl
Period
2006Q43.090428
2007Q11.676421
2007Q20.219586
2007Q31.570087
2007Q40.350137
2008Q10.438966
2008Q20.615457
2008Q30.119022
2008Q40.257887
2009Q10.129785
2009Q2-0.056094
2009Q3-0.066123
\n", + "
" + ], + "text/plain": [ + " infl\n", + "Period \n", + "2006Q4 3.090428\n", + "2007Q1 1.676421\n", + "2007Q2 0.219586\n", + "2007Q3 1.570087\n", + "2007Q4 0.350137\n", + "2008Q1 0.438966\n", + "2008Q2 0.615457\n", + "2008Q3 0.119022\n", + "2008Q4 0.257887\n", + "2009Q1 0.129785\n", + "2009Q2 -0.056094\n", + "2009Q3 -0.066123" + ] + }, + "execution_count": 13, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "pipe.fit(y=y_train, X=X_train, fh=fh)\n", + "result = pipe.predict(X=None, fh=y_test.index)\n", + "result" + ] + }, + { + "cell_type": "code", + "execution_count": 14, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "data": { + "text/plain": [ + "infl 20.103326\n", + "dtype: float64" + ] + }, + "execution_count": 14, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "((result - y_test) ** 2).mean()" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "#### Grid Search with graphical pipeline\n", + "\n", + "This pipeline has multiple parameters that might be tested to find the configurations. These parameters include:\n", + "* which forecaster should be used for which variable -> `MultiplexForecaster`\n", + "* what should be the hyperparameters of the forecaster\n", + "* which features should be used for the different forecasters -> Tune the edges of the graphical pipeline!\n", + "\n", + "\n", + "\n", + "Since we do forecasting, we use the ForecastingGridSearchCV." + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "1. Create blue print of the pipeline\n" + ] + }, + { + "cell_type": "code", + "execution_count": 15, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n" + ] + } + ], + "source": [ + "from sktime.forecasting.compose import MultiplexForecaster\n", + "\n", + "pipe = Pipeline()\n", + "sklearn_scaler = StandardScaler()\n", + "sktime_scaler = TabularToSeriesAdaptor(sklearn_scaler)\n", + "deseasonalizer = Deseasonalizer(sp=4)\n", + "\n", + "pipe = pipe.add_step(\n", + " sktime_scaler, name=\"scaler\", edges={\"X\": \"X__realgdp_realdpi_unemp\"}\n", + ")\n", + "pipe = pipe.add_step(\n", + " deseasonalizer, name=\"deseasonalizer\", edges={\"X\": \"X__realgdp_realdpi\"}\n", + ")\n", + "\n", + "pipe = pipe.add_step(\n", + " MultiplexForecaster(\n", + " [\n", + " (\n", + " \"ridge\",\n", + " make_reduction(Ridge(), windows_identical=False, window_length=5),\n", + " ),\n", + " (\n", + " \"lasso\",\n", + " make_reduction(Lasso(), windows_identical=False, window_length=5),\n", + " ),\n", + " ]\n", + " ),\n", + " name=\"forecaster_gdp\",\n", + " edges={\"y\": \"deseasonalizer__realgdp\"},\n", + ")\n", + "\n", + "pipe = pipe.add_step(\n", + " MultiplexForecaster(\n", + " [\n", + " (\n", + " \"ridge\",\n", + " make_reduction(Ridge(), windows_identical=False, window_length=5),\n", + " ),\n", + " (\n", + " \"lasso\",\n", + " make_reduction(Lasso(), windows_identical=False, window_length=5),\n", + " ),\n", + " ]\n", + " ),\n", + " name=\"forecaster_dpi\",\n", + " edges={\"y\": \"deseasonalizer__realdpi\"},\n", + ")\n", + "\n", + "pipe = pipe.add_step(\n", + " MultiplexForecaster(\n", + " [\n", + " (\n", + " \"ridge\",\n", + " make_reduction(Ridge(), windows_identical=False, window_length=5),\n", + " ),\n", + " (\n", + " \"lasso\",\n", + " make_reduction(Lasso(), windows_identical=False, window_length=5),\n", + " ),\n", + " ]\n", + " ),\n", + " name=\"forecaster_unemp\",\n", + " edges={\n", + " \"y\": \"scaler__unemp\",\n", + " \"X\": [\n", + " \"forecaster_gdp\",\n", + " \"forecaster_dpi\",\n", + " ],\n", + " },\n", + ")\n", + "\n", + "pipe = pipe.add_step(\n", + " MultiplexForecaster(\n", + " [\n", + " (\n", + " \"ridge\",\n", + " make_reduction(Ridge(), windows_identical=False, window_length=5),\n", + " ),\n", + " (\n", + " \"lasso\",\n", + " make_reduction(Lasso(), windows_identical=False, window_length=5),\n", + " ),\n", + " ]\n", + " ),\n", + " name=\"forecaster_inflation\",\n", + " edges={\"X\": [\"forecaster_dpi\", \"forecaster_unemp\"], \"y\": \"y\"},\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "\n", + "2. Specify the parameter grid:\n", + "\n", + "The keys of the dictionary are the parameters' in the pipeline, and the values specify which options should be tested.\n", + "Keys have the following structure: parameter of a step `__skobject__` and input edges of a step `__edges_`." + ] + }, + { + "cell_type": "code", + "execution_count": 16, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "param_grid = {\n", + " \"forecaster_inflation__skobject__selected_forecaster\": [\"ridge\", \"lasso\"],\n", + " \"forecaster_unemp__skobject__selected_forecaster\": [\"ridge\", \"lasso\"],\n", + " \"forecaster_dpi__skobject__selected_forecaster\": [\"ridge\", \"lasso\"],\n", + " \"forecaster_gdp__skobject__selected_forecaster\": [\"ridge\", \"lasso\"],\n", + " \"forecaster_inflation__edges__X\": [\n", + " [\"forecaster_unemp\"],\n", + " [\"forecaster_unemp\", \"forecaster_dpi\"],\n", + " ],\n", + " \"forecaster_unemp__edges__X\": [\n", + " [],\n", + " [\"forecaster_dpi\"],\n", + " [\"forecaster_gdp\", \"forecaster_dpi\"],\n", + " ],\n", + " \"deseasonalizer__edges__X\": [\"X__realgdp_realdpi\", \"scaler__realgdp_realdpi\"],\n", + "}" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "Initialise the gridsearch using pipeline, cross-validation strategy, scoring, and param_grid.\n" + ] + }, + { + "cell_type": "code", + "execution_count": 17, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "from sktime.forecasting.model_selection import (\n", + " ForecastingGridSearchCV,\n", + " SlidingWindowSplitter,\n", + ")\n", + "from sktime.performance_metrics.forecasting import mean_absolute_error\n", + "\n", + "gridcv = ForecastingGridSearchCV(\n", + " pipe,\n", + " cv=SlidingWindowSplitter(\n", + " window_length=len(X_train) - 20,\n", + " step_length=4,\n", + " fh=[1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12],\n", + " ),\n", + " scoring=mean_absolute_error,\n", + " param_grid=param_grid,\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "Call fit on the gridsearch object." + ] + }, + { + "cell_type": "code", + "execution_count": 18, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "/Users/benediktheidrich/code/sktime/sktime/forecasting/model_selection/_tune.py:201: UserWarning: in ForecastingGridSearchCV, n_jobs and pre_dispatch parameters are deprecated and will be removed in 0.27.0. Please use n_jobs and pre_dispatch directly in the backend_params argument instead.\n", + " warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.644e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.435e+05, tolerance: 9.351e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 1.934e+05, tolerance: 5.168e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.754e+01, tolerance: 1.702e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.038e+02, tolerance: 1.703e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.733e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.470e+05, tolerance: 9.857e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.170e+05, tolerance: 5.465e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.735e+01, tolerance: 1.701e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.193e+02, tolerance: 1.679e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 5.690e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.600e+05, tolerance: 1.035e+05\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.280e+05, tolerance: 5.757e+04\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.956e+01, tolerance: 1.697e-02\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 2.986e+02, tolerance: 1.665e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/step.py:217: FutureWarning: The behavior of pd.concat with len(keys) != len(objs) is deprecated. In a future version this will raise instead of truncating to the smaller of the two sequences\n", + " input_data[step_name] = pd.concat(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.997e+02, tolerance: 1.843e-01\n", + " model = cd_fast.enet_coordinate_descent(\n", + "/Users/benediktheidrich/code/sktime/sktime/pipeline/pipeline.py:160: UserWarning: This generalised graphical pipeline is experimental, with all the usual risks of edge features. For mature alternatives, use single-purpose pipelines and compositors, such as TransformedTargetForecaster, ForecastingPipeline, ClassificationPipeline, etc., see for instance notebooks 01_forecasting.ipynb and 02_classification.ipynb athttps://github.com/sktime/sktime/blob/main/examples/.\n", + " warnings.warn(\n", + "/Users/benediktheidrich/.pyenv/versions/3.11.9/lib/python3.11/site-packages/sklearn/linear_model/_coordinate_descent.py:628: ConvergenceWarning: Objective did not converge. You might want to increase the number of iterations, check the scale of the features or consider increasing regularisation. Duality gap: 3.997e+02, tolerance: 1.843e-01\n", + " model = cd_fast.enet_coordinate_descent(\n" + ] + }, + { + "data": { + "text/html": [ + "
ForecastingGridSearchCV(cv=SlidingWindowSplitter(fh=[1, 2, 3, 4, 5, 6, 7, 8, 9,\n",
+       "                                                     10, 11, 12],\n",
+       "                                                 step_length=4,\n",
+       "                                                 window_length=171),\n",
+       "                        forecaster=Pipeline(steps=[{'edges': {'X': 'X__realgdp_realdpi_unemp'},\n",
+       "                                                    'kwargs': {},\n",
+       "                                                    'method': None,\n",
+       "                                                    'name': 'scaler',\n",
+       "                                                    'skobject': TabularToSeriesAdaptor(transformer=StandardScaler())},\n",
+       "                                                   {'edges': {'X': 'X__realgdp_realdpi'},\n",
+       "                                                    'kwargs': {},\n",
+       "                                                    'method': Non...\n",
+       "                                    'forecaster_inflation__edges__X': [['forecaster_unemp'],\n",
+       "                                                                       ['forecaster_unemp',\n",
+       "                                                                        'forecaster_dpi']],\n",
+       "                                    'forecaster_inflation__skobject__selected_forecaster': ['ridge',\n",
+       "                                                                                            'lasso'],\n",
+       "                                    'forecaster_unemp__edges__X': [[],\n",
+       "                                                                   ['forecaster_dpi'],\n",
+       "                                                                   ['forecaster_gdp',\n",
+       "                                                                    'forecaster_dpi']],\n",
+       "                                    'forecaster_unemp__skobject__selected_forecaster': ['ridge',\n",
+       "                                                                                        'lasso']},\n",
+       "                        scoring=<function mean_absolute_error at 0x172c7e980>)
Please rerun this cell to show the HTML repr or trust the notebook.
" + ], + "text/plain": [ + "ForecastingGridSearchCV(cv=SlidingWindowSplitter(fh=[1, 2, 3, 4, 5, 6, 7, 8, 9,\n", + " 10, 11, 12],\n", + " step_length=4,\n", + " window_length=171),\n", + " forecaster=Pipeline(steps=[{'edges': {'X': 'X__realgdp_realdpi_unemp'},\n", + " 'kwargs': {},\n", + " 'method': None,\n", + " 'name': 'scaler',\n", + " 'skobject': TabularToSeriesAdaptor(transformer=StandardScaler())},\n", + " {'edges': {'X': 'X__realgdp_realdpi'},\n", + " 'kwargs': {},\n", + " 'method': Non...\n", + " 'forecaster_inflation__edges__X': [['forecaster_unemp'],\n", + " ['forecaster_unemp',\n", + " 'forecaster_dpi']],\n", + " 'forecaster_inflation__skobject__selected_forecaster': ['ridge',\n", + " 'lasso'],\n", + " 'forecaster_unemp__edges__X': [[],\n", + " ['forecaster_dpi'],\n", + " ['forecaster_gdp',\n", + " 'forecaster_dpi']],\n", + " 'forecaster_unemp__skobject__selected_forecaster': ['ridge',\n", + " 'lasso']},\n", + " scoring=)" + ] + }, + "execution_count": 18, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "gridcv.fit(y=y_train, X=X_train)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Examine the results of the gridsearch" + ] + }, + { + "cell_type": "code", + "execution_count": 19, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
mean_test__DynamicForecastingErrorMetricmean_fit_timemean_pred_timeparamsrank_test__DynamicForecastingErrorMetric
01.5393290.0756730.023929{'deseasonalizer__edges__X': 'X__realgdp_reald...107.5
11.7205650.0762080.025338{'deseasonalizer__edges__X': 'X__realgdp_reald...119.5
21.3943290.1418000.046452{'deseasonalizer__edges__X': 'X__realgdp_reald...97.5
31.9420510.1511810.045115{'deseasonalizer__edges__X': 'X__realgdp_reald...129.5
42.0337140.1604420.059711{'deseasonalizer__edges__X': 'X__realgdp_reald...136.0
..................
1871.3290790.0967610.037935{'deseasonalizer__edges__X': 'scaler__realgdp_...48.5
1881.3290790.1099970.040909{'deseasonalizer__edges__X': 'scaler__realgdp_...48.5
1891.3290790.1006590.047549{'deseasonalizer__edges__X': 'scaler__realgdp_...48.5
1901.3290790.1050450.055426{'deseasonalizer__edges__X': 'scaler__realgdp_...48.5
1911.3290790.1069060.039190{'deseasonalizer__edges__X': 'scaler__realgdp_...48.5
\n", + "

192 rows × 5 columns

\n", + "
" + ], + "text/plain": [ + " mean_test__DynamicForecastingErrorMetric mean_fit_time mean_pred_time \\\n", + "0 1.539329 0.075673 0.023929 \n", + "1 1.720565 0.076208 0.025338 \n", + "2 1.394329 0.141800 0.046452 \n", + "3 1.942051 0.151181 0.045115 \n", + "4 2.033714 0.160442 0.059711 \n", + ".. ... ... ... \n", + "187 1.329079 0.096761 0.037935 \n", + "188 1.329079 0.109997 0.040909 \n", + "189 1.329079 0.100659 0.047549 \n", + "190 1.329079 0.105045 0.055426 \n", + "191 1.329079 0.106906 0.039190 \n", + "\n", + " params \\\n", + "0 {'deseasonalizer__edges__X': 'X__realgdp_reald... \n", + "1 {'deseasonalizer__edges__X': 'X__realgdp_reald... \n", + "2 {'deseasonalizer__edges__X': 'X__realgdp_reald... \n", + "3 {'deseasonalizer__edges__X': 'X__realgdp_reald... \n", + "4 {'deseasonalizer__edges__X': 'X__realgdp_reald... \n", + ".. ... \n", + "187 {'deseasonalizer__edges__X': 'scaler__realgdp_... \n", + "188 {'deseasonalizer__edges__X': 'scaler__realgdp_... \n", + "189 {'deseasonalizer__edges__X': 'scaler__realgdp_... \n", + "190 {'deseasonalizer__edges__X': 'scaler__realgdp_... \n", + "191 {'deseasonalizer__edges__X': 'scaler__realgdp_... \n", + "\n", + " rank_test__DynamicForecastingErrorMetric \n", + "0 107.5 \n", + "1 119.5 \n", + "2 97.5 \n", + "3 129.5 \n", + "4 136.0 \n", + ".. ... \n", + "187 48.5 \n", + "188 48.5 \n", + "189 48.5 \n", + "190 48.5 \n", + "191 48.5 \n", + "\n", + "[192 rows x 5 columns]" + ] + }, + "execution_count": 19, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "gridcv.cv_results_" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "Using the fitted grid search to make a prediction with the best hyperparameters" + ] + }, + { + "cell_type": "code", + "execution_count": 20, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
infl
Period
2006Q42.188182
2007Q12.124281
2007Q21.045280
2007Q31.857716
2007Q41.790664
2008Q11.649457
2008Q21.874361
2008Q31.855627
2008Q41.858207
2009Q11.909693
2009Q21.905106
2009Q31.910452
\n", + "
" + ], + "text/plain": [ + " infl\n", + "Period \n", + "2006Q4 2.188182\n", + "2007Q1 2.124281\n", + "2007Q2 1.045280\n", + "2007Q3 1.857716\n", + "2007Q4 1.790664\n", + "2008Q1 1.649457\n", + "2008Q2 1.874361\n", + "2008Q3 1.855627\n", + "2008Q4 1.858207\n", + "2009Q1 1.909693\n", + "2009Q2 1.905106\n", + "2009Q3 1.910452" + ] + }, + "execution_count": 20, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "result = gridcv.predict(X=None, fh=y_test.index)\n", + "result" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "### How to implement a bit simpler version of the pipeline above by nesting sequential pipelines\n", + "* Simplifcation: The forecasting of the unemployment rate is not dependent on the GDP and DPI.\n", + "\n", + "\n" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "Create sequential pipelines for forecasting the GDP, DPI and unemployment rate." + ] + }, + { + "cell_type": "code", + "execution_count": 21, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "from sktime.forecasting.compose import ColumnEnsembleForecaster, ForecastX\n", + "from sktime.transformations.series.subset import ColumnSelect\n", + "\n", + "forecasting_pipeline_gdp = (\n", + " ColumnSelect([\"realgdp\"]) # To train the forecaster only on the realgdp column\n", + " * Deseasonalizer()\n", + " * MultiplexForecaster(\n", + " [\n", + " (\n", + " \"ridge\",\n", + " make_reduction(Ridge(), windows_identical=False, window_length=5),\n", + " ),\n", + " (\n", + " \"lasso\",\n", + " make_reduction(Lasso(), windows_identical=False, window_length=5),\n", + " ),\n", + " ]\n", + " )\n", + ")\n", + "forecasting_pipeline_dpi = (\n", + " ColumnSelect([\"realdpi\"])\n", + " * Deseasonalizer()\n", + " * MultiplexForecaster(\n", + " [\n", + " (\n", + " \"ridge\",\n", + " make_reduction(Ridge(), windows_identical=False, window_length=5),\n", + " ),\n", + " (\n", + " \"lasso\",\n", + " make_reduction(Lasso(), windows_identical=False, window_length=5),\n", + " ),\n", + " ]\n", + " )\n", + ")\n", + "\n", + "forecasting_pipeline_unemp = (\n", + " ColumnSelect([\"unemp\"])\n", + " * Deseasonalizer()\n", + " * MultiplexForecaster(\n", + " [\n", + " (\n", + " \"ridge\",\n", + " make_reduction(Ridge(), windows_identical=False, window_length=5),\n", + " ),\n", + " (\n", + " \"lasso\",\n", + " make_reduction(Lasso(), windows_identical=False, window_length=5),\n", + " ),\n", + " ]\n", + " )\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "Use ColunmEnsembleForecaster to combine the forecasts of the DPI, GDP, UNEMP. (Union of forecasts)" + ] + }, + { + "cell_type": "code", + "execution_count": 22, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "input_inflation_forecast = ColumnEnsembleForecaster(\n", + " [\n", + " (\"realdpi\", forecasting_pipeline_dpi, \"realdpi\"),\n", + " (\"realgdp\", forecasting_pipeline_gdp, \"realgdp\"),\n", + " (\"unemp\", forecasting_pipeline_unemp, \"unemp\"),\n", + " ]\n", + ")" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "Create the inflation forecaster." + ] + }, + { + "cell_type": "code", + "execution_count": 23, + "metadata": { + "collapsed": false + }, + "outputs": [], + "source": [ + "inflation_forecast = ForecastX(\n", + " MultiplexForecaster(\n", + " [\n", + " (\n", + " \"ridge\",\n", + " make_reduction(Ridge(), windows_identical=False, window_length=5),\n", + " ),\n", + " (\n", + " \"lasso\",\n", + " make_reduction(Lasso(), windows_identical=False, window_length=5),\n", + " ),\n", + " ]\n", + " ),\n", + " input_inflation_forecast,\n", + ")" + ] + }, + { + "cell_type": "code", + "execution_count": 24, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "data": { + "text/html": [ + "
ForecastX(forecaster_X=ColumnEnsembleForecaster(forecasters=[('realdpi',\n",
+       "                                                              TransformedTargetForecaster(steps=[ColumnSelect(columns=['realdpi']),\n",
+       "                                                                                                 Deseasonalizer(),\n",
+       "                                                                                                 MultiplexForecaster(forecasters=[('ridge',\n",
+       "                                                                                                                                   RecursiveTabularRegressionForecaster(estimator=Ridge(),\n",
+       "                                                                                                                                                                        window_length=5)),\n",
+       "                                                                                                                                  ('lasso',\n",
+       "                                                                                                                                   RecursiveTabularRegressionForecaster(estimator=Lasso(),\n",
+       "                                                                                                                                                                        window_length...\n",
+       "                                                                                                                                   RecursiveTabularRegressionForecaster(estimator=Ridge(),\n",
+       "                                                                                                                                                                        window_length=5)),\n",
+       "                                                                                                                                  ('lasso',\n",
+       "                                                                                                                                   RecursiveTabularRegressionForecaster(estimator=Lasso(),\n",
+       "                                                                                                                                                                        window_length=5))])]),\n",
+       "                                                              'unemp')]),\n",
+       "          forecaster_y=MultiplexForecaster(forecasters=[('ridge',\n",
+       "                                                         RecursiveTabularRegressionForecaster(estimator=Ridge(),\n",
+       "                                                                                              window_length=5)),\n",
+       "                                                        ('lasso',\n",
+       "                                                         RecursiveTabularRegressionForecaster(estimator=Lasso(),\n",
+       "                                                                                              window_length=5))]))
Please rerun this cell to show the HTML repr or trust the notebook.
" + ], + "text/plain": [ + "ForecastX(forecaster_X=ColumnEnsembleForecaster(forecasters=[('realdpi',\n", + " TransformedTargetForecaster(steps=[ColumnSelect(columns=['realdpi']),\n", + " Deseasonalizer(),\n", + " MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length...\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])]),\n", + " 'unemp')]),\n", + " forecaster_y=MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))]))" + ] + }, + "execution_count": 24, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "inflation_forecast.fit(y=y_train, X=X_train, fh=fh)" + ] + }, + { + "cell_type": "code", + "execution_count": 25, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
infl
2006Q43.979318
2007Q12.347512
2007Q21.443598
2007Q33.914533
2007Q42.533117
2008Q13.278010
2008Q23.861517
2008Q33.487510
2008Q44.195074
2009Q14.294984
2009Q24.433578
2009Q34.858610
\n", + "
" + ], + "text/plain": [ + " infl\n", + "2006Q4 3.979318\n", + "2007Q1 2.347512\n", + "2007Q2 1.443598\n", + "2007Q3 3.914533\n", + "2007Q4 2.533117\n", + "2008Q1 3.278010\n", + "2008Q2 3.861517\n", + "2008Q3 3.487510\n", + "2008Q4 4.195074\n", + "2009Q1 4.294984\n", + "2009Q2 4.433578\n", + "2009Q3 4.858610" + ] + }, + "execution_count": 25, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "inflation_forecast.predict()" + ] + }, + { + "cell_type": "code", + "execution_count": 26, + "metadata": { + "collapsed": false + }, + "outputs": [ + { + "data": { + "text/plain": [ + "{'behaviour': 'update',\n", + " 'columns': None,\n", + " 'fh_X': None,\n", + " 'fit_behaviour': 'use_actual',\n", + " 'forecaster_X': ColumnEnsembleForecaster(forecasters=[('realdpi',\n", + " TransformedTargetForecaster(steps=[ColumnSelect(columns=['realdpi']),\n", + " Deseasonalizer(),\n", + " MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])]),\n", + " 'realdpi'),\n", + " ('r...\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])]),\n", + " 'realgdp'),\n", + " ('unemp',\n", + " TransformedTargetForecaster(steps=[ColumnSelect(columns=['unemp']),\n", + " Deseasonalizer(),\n", + " MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])]),\n", + " 'unemp')]),\n", + " 'forecaster_y': MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))]),\n", + " 'forecaster_X__forecasters': [('realdpi',\n", + " TransformedTargetForecaster(steps=[ColumnSelect(columns=['realdpi']),\n", + " Deseasonalizer(),\n", + " MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])]),\n", + " 'realdpi'),\n", + " ('realgdp',\n", + " TransformedTargetForecaster(steps=[ColumnSelect(columns=['realgdp']),\n", + " Deseasonalizer(),\n", + " MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])]),\n", + " 'realgdp'),\n", + " ('unemp',\n", + " TransformedTargetForecaster(steps=[ColumnSelect(columns=['unemp']),\n", + " Deseasonalizer(),\n", + " MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])]),\n", + " 'unemp')],\n", + " 'forecaster_X__realdpi': TransformedTargetForecaster(steps=[ColumnSelect(columns=['realdpi']),\n", + " Deseasonalizer(),\n", + " MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])]),\n", + " 'forecaster_X__realgdp': TransformedTargetForecaster(steps=[ColumnSelect(columns=['realgdp']),\n", + " Deseasonalizer(),\n", + " MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])]),\n", + " 'forecaster_X__unemp': TransformedTargetForecaster(steps=[ColumnSelect(columns=['unemp']),\n", + " Deseasonalizer(),\n", + " MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])]),\n", + " 'forecaster_X__realdpi__steps': [ColumnSelect(columns=['realdpi']),\n", + " Deseasonalizer(),\n", + " MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])],\n", + " 'forecaster_X__realdpi__ColumnSelect': ColumnSelect(columns=['realdpi']),\n", + " 'forecaster_X__realdpi__Deseasonalizer': Deseasonalizer(),\n", + " 'forecaster_X__realdpi__MultiplexForecaster': MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))]),\n", + " 'forecaster_X__realdpi__ColumnSelect__columns': ['realdpi'],\n", + " 'forecaster_X__realdpi__ColumnSelect__index_treatment': 'remove',\n", + " 'forecaster_X__realdpi__ColumnSelect__integer_treatment': 'col',\n", + " 'forecaster_X__realdpi__Deseasonalizer__model': 'additive',\n", + " 'forecaster_X__realdpi__Deseasonalizer__sp': 1,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__forecasters': [('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(), window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(), window_length=5))],\n", + " 'forecaster_X__realdpi__MultiplexForecaster__selected_forecaster': None,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge': RecursiveTabularRegressionForecaster(estimator=Ridge(), window_length=5),\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso': RecursiveTabularRegressionForecaster(estimator=Lasso(), window_length=5),\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge__estimator': Ridge(),\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge__pooling': 'local',\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge__transformers': None,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge__window_length': 5,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge__estimator__alpha': 1.0,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge__estimator__copy_X': True,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge__estimator__fit_intercept': True,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge__estimator__max_iter': None,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge__estimator__positive': False,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge__estimator__random_state': None,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge__estimator__solver': 'auto',\n", + " 'forecaster_X__realdpi__MultiplexForecaster__ridge__estimator__tol': 0.0001,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__estimator': Lasso(),\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__pooling': 'local',\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__transformers': None,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__window_length': 5,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__estimator__alpha': 1.0,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__estimator__copy_X': True,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__estimator__fit_intercept': True,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__estimator__max_iter': 1000,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__estimator__positive': False,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__estimator__precompute': False,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__estimator__random_state': None,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__estimator__selection': 'cyclic',\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__estimator__tol': 0.0001,\n", + " 'forecaster_X__realdpi__MultiplexForecaster__lasso__estimator__warm_start': False,\n", + " 'forecaster_X__realgdp__steps': [ColumnSelect(columns=['realgdp']),\n", + " Deseasonalizer(),\n", + " MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])],\n", + " 'forecaster_X__realgdp__ColumnSelect': ColumnSelect(columns=['realgdp']),\n", + " 'forecaster_X__realgdp__Deseasonalizer': Deseasonalizer(),\n", + " 'forecaster_X__realgdp__MultiplexForecaster': MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))]),\n", + " 'forecaster_X__realgdp__ColumnSelect__columns': ['realgdp'],\n", + " 'forecaster_X__realgdp__ColumnSelect__index_treatment': 'remove',\n", + " 'forecaster_X__realgdp__ColumnSelect__integer_treatment': 'col',\n", + " 'forecaster_X__realgdp__Deseasonalizer__model': 'additive',\n", + " 'forecaster_X__realgdp__Deseasonalizer__sp': 1,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__forecasters': [('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(), window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(), window_length=5))],\n", + " 'forecaster_X__realgdp__MultiplexForecaster__selected_forecaster': None,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge': RecursiveTabularRegressionForecaster(estimator=Ridge(), window_length=5),\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso': RecursiveTabularRegressionForecaster(estimator=Lasso(), window_length=5),\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge__estimator': Ridge(),\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge__pooling': 'local',\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge__transformers': None,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge__window_length': 5,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge__estimator__alpha': 1.0,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge__estimator__copy_X': True,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge__estimator__fit_intercept': True,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge__estimator__max_iter': None,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge__estimator__positive': False,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge__estimator__random_state': None,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge__estimator__solver': 'auto',\n", + " 'forecaster_X__realgdp__MultiplexForecaster__ridge__estimator__tol': 0.0001,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__estimator': Lasso(),\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__pooling': 'local',\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__transformers': None,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__window_length': 5,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__estimator__alpha': 1.0,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__estimator__copy_X': True,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__estimator__fit_intercept': True,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__estimator__max_iter': 1000,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__estimator__positive': False,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__estimator__precompute': False,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__estimator__random_state': None,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__estimator__selection': 'cyclic',\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__estimator__tol': 0.0001,\n", + " 'forecaster_X__realgdp__MultiplexForecaster__lasso__estimator__warm_start': False,\n", + " 'forecaster_X__unemp__steps': [ColumnSelect(columns=['unemp']),\n", + " Deseasonalizer(),\n", + " MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))])],\n", + " 'forecaster_X__unemp__ColumnSelect': ColumnSelect(columns=['unemp']),\n", + " 'forecaster_X__unemp__Deseasonalizer': Deseasonalizer(),\n", + " 'forecaster_X__unemp__MultiplexForecaster': MultiplexForecaster(forecasters=[('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(),\n", + " window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(),\n", + " window_length=5))]),\n", + " 'forecaster_X__unemp__ColumnSelect__columns': ['unemp'],\n", + " 'forecaster_X__unemp__ColumnSelect__index_treatment': 'remove',\n", + " 'forecaster_X__unemp__ColumnSelect__integer_treatment': 'col',\n", + " 'forecaster_X__unemp__Deseasonalizer__model': 'additive',\n", + " 'forecaster_X__unemp__Deseasonalizer__sp': 1,\n", + " 'forecaster_X__unemp__MultiplexForecaster__forecasters': [('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(), window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(), window_length=5))],\n", + " 'forecaster_X__unemp__MultiplexForecaster__selected_forecaster': None,\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge': RecursiveTabularRegressionForecaster(estimator=Ridge(), window_length=5),\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso': RecursiveTabularRegressionForecaster(estimator=Lasso(), window_length=5),\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge__estimator': Ridge(),\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge__pooling': 'local',\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge__transformers': None,\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge__window_length': 5,\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge__estimator__alpha': 1.0,\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge__estimator__copy_X': True,\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge__estimator__fit_intercept': True,\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge__estimator__max_iter': None,\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge__estimator__positive': False,\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge__estimator__random_state': None,\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge__estimator__solver': 'auto',\n", + " 'forecaster_X__unemp__MultiplexForecaster__ridge__estimator__tol': 0.0001,\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__estimator': Lasso(),\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__pooling': 'local',\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__transformers': None,\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__window_length': 5,\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__estimator__alpha': 1.0,\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__estimator__copy_X': True,\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__estimator__fit_intercept': True,\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__estimator__max_iter': 1000,\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__estimator__positive': False,\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__estimator__precompute': False,\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__estimator__random_state': None,\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__estimator__selection': 'cyclic',\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__estimator__tol': 0.0001,\n", + " 'forecaster_X__unemp__MultiplexForecaster__lasso__estimator__warm_start': False,\n", + " 'forecaster_y__forecasters': [('ridge',\n", + " RecursiveTabularRegressionForecaster(estimator=Ridge(), window_length=5)),\n", + " ('lasso',\n", + " RecursiveTabularRegressionForecaster(estimator=Lasso(), window_length=5))],\n", + " 'forecaster_y__selected_forecaster': None,\n", + " 'forecaster_y__ridge': RecursiveTabularRegressionForecaster(estimator=Ridge(), window_length=5),\n", + " 'forecaster_y__lasso': RecursiveTabularRegressionForecaster(estimator=Lasso(), window_length=5),\n", + " 'forecaster_y__ridge__estimator': Ridge(),\n", + " 'forecaster_y__ridge__pooling': 'local',\n", + " 'forecaster_y__ridge__transformers': None,\n", + " 'forecaster_y__ridge__window_length': 5,\n", + " 'forecaster_y__ridge__estimator__alpha': 1.0,\n", + " 'forecaster_y__ridge__estimator__copy_X': True,\n", + " 'forecaster_y__ridge__estimator__fit_intercept': True,\n", + " 'forecaster_y__ridge__estimator__max_iter': None,\n", + " 'forecaster_y__ridge__estimator__positive': False,\n", + " 'forecaster_y__ridge__estimator__random_state': None,\n", + " 'forecaster_y__ridge__estimator__solver': 'auto',\n", + " 'forecaster_y__ridge__estimator__tol': 0.0001,\n", + " 'forecaster_y__lasso__estimator': Lasso(),\n", + " 'forecaster_y__lasso__pooling': 'local',\n", + " 'forecaster_y__lasso__transformers': None,\n", + " 'forecaster_y__lasso__window_length': 5,\n", + " 'forecaster_y__lasso__estimator__alpha': 1.0,\n", + " 'forecaster_y__lasso__estimator__copy_X': True,\n", + " 'forecaster_y__lasso__estimator__fit_intercept': True,\n", + " 'forecaster_y__lasso__estimator__max_iter': 1000,\n", + " 'forecaster_y__lasso__estimator__positive': False,\n", + " 'forecaster_y__lasso__estimator__precompute': False,\n", + " 'forecaster_y__lasso__estimator__random_state': None,\n", + " 'forecaster_y__lasso__estimator__selection': 'cyclic',\n", + " 'forecaster_y__lasso__estimator__tol': 0.0001,\n", + " 'forecaster_y__lasso__estimator__warm_start': False}" + ] + }, + "execution_count": 26, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "inflation_forecast.get_params(True)" + ] + }, + { + "cell_type": "markdown", + "metadata": { + "collapsed": false + }, + "source": [ + "# Comparison graphical pipeline with nesting of sequential pipelines\n", + "\n", + "### Advantages of graphical pipelines\n", + "* Enable an easy implementation of complex pipelines\n", + " * By nesting sequential pipelines, even a simplified version of the graphical pipeline is very complicat to implement.\n", + " * By nesting sequential pipelines, some graphical pipelines are not possible to implement (e.g., the example with coupled ForecastX).\n", + "* Preprocessing steps can not be shared between the different forecasters.\n", + "* The parameter structure can be very complex for the sequential pipelines.\n", + "* In a complex scenario, how would you fine-tune the edges?\n", + "\n", + "### Advantages of sequential pipelines\n", + "* Constructing simple pipelines is very easy.\n", + "* Inverse operations are automatically applied.\n", + "* This is a mature feature compared to the experimental graphical pipeline.\n", + "\n", + "\n", + "### When to use what? \n", + "* If your pipeline does not need much of nested pipelines and is mainly sequential, you should probably stick with the standard pipeline implementation.\n", + "* If your pipeline should represent a complex scenario with multiple forecasters, that are influencing other ones, you might want to use the graphical pipeline since it makes it easier to write the codel\n" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python 3", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.11.9" + } + }, + "nbformat": 4, + "nbformat_minor": 0 +} diff --git a/examples/img/classification_pipeline.png b/examples/img/classification_pipeline.png new file mode 100644 index 0000000000000000000000000000000000000000..e94e32e47da944921e7971ef85987b609334221f GIT binary patch literal 22850 zcmaI81wfR~)-aBwlyrAD0#ef5CAE|wCEbmrq;!LHNW;>xz|xI$DXBE9bV%=i(fhmK zec$_i|3{c-=H$$oGiT0>kABinlgC6OMT3Kb!&Fp|(S(CTdJ6~lEc7`7th5A4c)}3R z+%)B-;D8h4`>+P0m87a999&&I`n@R<92|VBhKjZ<3{Jr#I|T>#2nQ$Wk_~Gv!@*I$ zl804b4-5!<#2wOM<m zN8rD)w$7f2C{GrG_4XM&{3bH$Uz7~&1qV0#r{v(``P&~@d47ElgTaWY1=ZUB5etK? zL10SYR^i~1!J9C$@;{6w%2u#?Q}@Ud3lvJkA_!|f|AT@_fIYCL4Ka;`QzooVNlgoD z!k+Q(+XM`+L~WB`^?%z?$TIpb_AjIG``CY@j^W@`G_+yFFoLHS9GswO_+MgJJu$No zR{op#-|+DtXMJ?suqGBZw#XCH)AJYmx8(W*fiYGzg8!oE;o#cl5C2W(RJw$Oiou~~ z2ZJ>H;r-h;ctj+agysD!7+g$J7FI8X)!^VT;NT3i=3y1pKSP7T>BRw|F|hf3(z|gRvx@8IB2{y^roFJNzCHepx^BmVp3me#Zj z&_=-;yY8B5T5!;cA!r{IYV>?^4B85{wt!+kUz`nrLY?i&iI1?M>&wvS2&g+fHI#T4 zx|tjgn=C(eSiwvUJ%&QE(xFgZE-3Tw^P)T`RFDe_y{m+lLM5OLH7ZaA2`IGPvjs|f z1l7LNfGO#Mse$TgAdW$yLr{x%Few;_vy)JJ>v#B2jO8>abTI-N0(F1s3?;|kT#r|Z zhUtPrz3Cyl>B-O_Zz%gY6ndPO1>;+R3I_2)@1Vb7%8H;+5h%0<3awOtVW2W1VbB(6 z15`x@+VvRL4z+@o>EUTt`3Kg+hK%a0pzj6;M}+hA`)ri)TpSKghgwlaQrp}35F8ys zJ`dhqM0H)@`W(z{#ANgVJ5qQI^F_jQ9(|c#I@CJ_?5MwhNMC)Wk5y#Ir3Y}ZYid0) z*~7bKlv^=2iiJ(hJQ!@^j(KYNtHkoC@Niu{gLG1*+YIy~Krq9lv@u`(mEvgTYFjUELOz!-Y!i z1X~XbJmELhQ<@5R%(l&8tA%`LLPPsb;5OsW@nub;BtepSC)rbQ!8o6Bsyryx2NiG% zCb4r$`QdwTi4(;-ylfJ;*uH_WT+Kw8tF*Z21zlqX#z(m^bZx~*2%GEo(}TOR4d;Sw z%}rC;PB^rQh7U?uExgg9GA&J?f$Z>9x9#Qq`K^PxpVRq!{%B!ON{br0ibOuQP;d$h zo1?4ih*1Ud(k7C694@p8pM;Xz`g7{-evFTQ1|>VijZRc7A#atlz=w> zBPkH=_n$!N+|ScPFFfJagWh`{(S6mkLSdAjsfVu;OrjV;gUsvh)@e`Dax7x;Oscfs-L-3}FRiWfB+gkmbQn7%kN{4->m!w5N%ckiwx0KLh{S)=bay3+D~Q&?sWZ>s#csDh;;+iL;So}L)sg>OaTc(l`wWaxAinsE`sl2Hx z>Q>r%oICys0@k`78cc1q4hrB}EDdgP83Pta6bK?l+VnYSceDiOI$wl?YnOvQ!*dB`%7kqnbl~h8?N`^Dl$wbJyFcsa zWC!e8DiF!W^meKZh^TAI&g$ts+Mw5kRGyoCeGSJ;c*Pu$Aelu`#pS!)gEokkMWFkw zs(aFrj4Rm`9|LjZoyx(~8$Qi`!cOxa*B=8BJWz%mv0NdBnj5N}i@}evpi3o7zq!&ztU!VH}IwV z6~oY5RS`CbQ$IsCkGa_Yc0DNa z>>Ig5w+hLOp()L!LRD=nJ+u9J*fXyDuuEG?ld&YbmydkYv}hmj*U@%RPwcMj!L^sZ z6X4p|(zZ-Ah*T!#TZW5~uJ>e4!aXm&B!(#1!%Qk2>8GmHXXBcSTm!UygpnhC;Dypk z#`usdaJtz_z;>!v+>^qF%Y0f{SGIosv+2Q`=G9zCM9utt?SzXE)cCQ{?UG=LhKbc^ zd@Xm=V(0rn)UefA4ZV*SvD$TE4$8@7_bj7X?@d3zbENUVD-nVi^%#A_T*bxcgnN%D zE@{fNt#u?Qp}ALA3wDz`5=dx&rmk>$ZFh`&oDf5778w}lO7WA5iFBuzQY;kLO8*1$7Xd^1ScEwunmZYtR}E;IbJ%5!~HLKQ7jGOS);vDsQY2}&@c zu$~JqVe{s+!zqT#IDfDYGBTUJ?hh*g(y!*k;cqZ-JcVK=@Le8Two6DE{2T;X8u_bg z$CV;^bgJByd((e!E1Q+r8r86$5I{X1^HJwV-_I%@s%5{%Yb#|AS}X}PsCz?_R$>NV zJ2$9v?BMCr`3$g88qftrqhvN#P-(dwGtx?ViHA9-IsiV4E3Bt#CxYlwd=b~hB9MJM%;&xl7+bSPNu$SljOUad$rzNAZ|h*SF+{622;INm6@>AwW@^><*{)8?;|SOyf)7EQprJ=Ce8-G^jk?mk z8tHM*6oTKJ--hBBIXWymKX$YQ5gOnoNE>n);tH`gzv8ZbXWq9wb@W{-l%gq>dLzCy zU~))8*pq%eXt;o51i}~*m1>X@h4G+ria?aT(q{|{^mXkPxYP8-;yVZas0h_D?gN~s z;N3Dk)hPiT7EG1%QVfVpVYyjqI>@)CM@9A0c{Tbe&eq+8*rmMr1RlyE8DH~$_4Pg zeYZ2)Ey6>Wb~$Q1WBKNVW1(@K?X+)Dm6+PRB3y%|E}NYh=}!%fw??54`sdn)(FblyETqgfk+E&VdS>gbWT;knXQ}^?ZKu?ns_w zp8?}ba@%ABxgr^Uf)9O_W#h`+W)})|X2*W$O*i;*F}x4Dzs3a$_^Fa5`%1M2gg`>G zFOD`I-G``VArO&c{)ux`^r!2Ch73CZ2S0>t~q*ZQhKyhZ1_15MfHSZD_6AxHf z(6hAd(W!|0fb0CFdfx56Zp7mUp-olg0m3nNEBU&r0{9Hghe0drO{yU+SgGYDDUVq1 z{)xOjXU*?v^1(&6wOoQ!Q=f~`W0p;)!z#FGxp_NF(nuOxU$t7l2A`h+7RRw48rlf{9rdpGT|EYHtn> z@8+_t$I0Opab9JgnP$xkDj41~PsW=uCG%R9b9G%3C!cKiQFf|7?-hGS3qi3|!mkkx zl|jgU$rs+zH1zuBhtqz78uF*I_UwonV}|CiQAP@^6T>f{klE3?(bp9x+(+dq2RfUn z$0sNbcsaP#sd(6L8brDbPww((v2F~yWeUkJN2!)^jf&gALb5dHrqfeML0_x1*B<@GW~oWYWnxN* zLuT7VH%~R^jb(B<&Nz!d2yPGCDi%z>W*2&W$HnkK*H%#l;*m=gQpY>t|8TF#N2?W1 z(9$#o&h@Pt;XEl{XVgfJeV1uU9G=L3@|~UFFfhhoM`OV+~wH_|W=_@yOHWqj=i2d74;S37CYG zszSeVQH^)5a#wXxbCs@~vdvt&IF z+3(BA6*~eOg9-L0jf4oY6MQYVZNW!BN&H5E0I$Zic2jiGd5AG?$yF^I1-Gksd;749 zVlQ43*l3>5sdQo_1#FDv<*V1}25uYl#iUO4EWPkO{05S40dCMi_y_|->oH$Z99)i} z2;Z?C)KCPbv>N-^;5%9GYt(3$Yd0P4Z6K`2Tm|d>Diayexnn##fS(9vc=oyx!HuDy z2+RFT>3a41_hsetlp{Byz@=#V(h2%?zg0SkD$!(Vu;W+9>$k`#yp6hsr*4m|fWr^O z)?<+im5u7Z)v8XW7lPkj)JZx3Yza?J`t8TFa?4dUTr&-u+2#o8q+WN8=#ZWAk7E}Z z)!eIMY9AqFHB}UCWhg(^%zK_y`g9_s-0h`Zm?PI_}YVRMcS;jEYThoSd`qr|}P zn2sUr9H+w4VH1!7Mz>bu8|c|1=rwR{GY&R+puUN~ff1U3_K|YwjP&7povZu*D)vZ;tk z=LN`+L6MbM`HJ68YC@ONc%5VM4!?z0x(g;1Pu#vyO_%ir!0+Vu5eg{lm!gXvi4OFH zq}kV>n9p~|G=`cqux0P}sJXdlw=aXu!8a1p5;~ilPwS537Z>ee@kJBU6O+=(7foLk z1E8OEE56cWyaC@j#kZ05JqrLgbJZ1AS9xA&)`;@pU*5vmUi95@6(%-cF2&*i0nlQj z7IQ(g%0=UA3fqf?^>;FcRmPhxrw?T_dFC=iRmSaS#I_eXzqx5&=+;npUqzJ77#k0l z>Vf6X@og`vv$zZ8ksFL%O6XAaKDcszz}Hz?%(_8tG)_GD{CIGwIk(s1*d4j`jtw@+ zPYZ$u>WgVibt58ofljLk?`n;*qa+2qyrs3h03Dmjd>Sb=cDh&}y+M{s$^8nsM3~#d zTknpXrtb57INC0oA>uYm87?u-laDn%L6eIiw^@W>u1lrH9SK{6;X^Jx=JtxxP5YUX zmB7d^lbZnNw-<&ri-mGs-nz(`Qw7Fkz|f{z^C97~nbO2moWf+|cu(NsD=V=;auD&j zdj()|{}C8>$X=uc=JZOaoVT+QdUVtWGnQz9kNUjmVZ+}qTU65nd-lU3T$OP{&wss~ z!NRD+Jx}{B>?bA|5%V|kyJvmO*pb}Gdfi-Tp`notY82ey_v6>s*BM1vKB9K=gCUf} z+f9WzE;ir~9VRPMO>GBTm!;1x70k?iDaaQr&pG0Rz2fkInwCwcADoMmr$V&U))dFt z&oENWMV53HN>{j**u%B@5Aabl%e8lgET(B)Aorj{{{3XB4FM|UpM-47=$rxjDm5V? z2dRna&!ZJuz9;gt6d#-HjKB3_@VTt%kk8(In?6_Z#k+$29)FqoAaI=dPnr+9P}q`CKSTqTG0pYPE^KrPep@TO9-HTtaGP>%cta zC0k%?LjNK`I&y`Nb+83Yz&8Jg6!R7P^9XV^dSbQCyLIg)NU9;K{rD2)^ZnH4JW6n( zB;#(UNBscFxYcXVM*4 z3BhONBEh>9M0^^{Q{{PlopG4C)RLClxxXhe*Qs?;&Sl3|0?cc19vs?r zC=Q-4A%w%Wu0KC=ben$nRS~~gW<>h})Z8!C#(aLyAr%8+RDF&*uiedXoA~ESSK^y1MOflz&~zIf24DX4mZ~U%Zbv z?>GtMZK?)x;bEivVX7pY>!M6lG#OG~wF2n7+4t3#_p|9DGA{Z>Db+@16ng2SKS`X2 z_96<*$&~zhz^Oa^X6Gi8(nP;ZgCoNjC8cdLM~m-(%i$fUk)fBY2A7@VX5S+0)~!0? zLg+|0BpRUtpm=JQltfXeSfeFGOKMZr$Q1X!(lKgIN&l09^nkAD#CaIDQmk-XLNq-N zZpq@aVj!SpGz*XDono7op`oKuv+Iu_Ro@)1a~d%2vMT~UPBGwdjt$!bj2rE1sOR_1 zWR>9lXxO{ftL?xOtYBn6U*tDq0_Ltx?^UTmh&Divj?x|xzqRn0XYn++U8T1+*D}-k zS_%X#of+ytc^pz}+%?TQ+LV;AGcOMYtsO}XytFKQvd0U_O!Q1jjjn;2x!%=8)p2ir zDtt6T1N);4lzbF4;Hykt6yMRU%p!+yv*z>rmsHR|1i&xED(Kc`9tuY(G|X1D6r^HM z3AxT9?yZ`J)(@-)hg*?THHw!1tU@}9skm{HsueC@MEKD^@6dj_Ny;4ss4v0z?h9Q< zd{hUNVP=8b^@|h)HJaVgfLqbmN~d^^GNYL4L#BK8dD;R! z!CsT+$3c#|P>3%>L0<8;Axt^4;-yf_uayA$4`1R{18V}BX6?`qYHZ%YhJp*Z{Nk3Z zaHT+%l}qK*R)hl1nRZ@XT)=g&3hr}Q9EYxWp2i4zXHs(8{!Lg~MxsH79e_TT8SPH#AvpbW0Eg=UA zmJ&brqBigvx6(}E&F?_}7a7egTRu58om5TFs7?^4cY`vX!j+17K~02p&K=U=g|xKb zO~NRD_tjTSPIt-5L#)wfm%v6*yCsPZ&=R)J-Xt*CdezBfuuO(GA{n9F9>IQc<%bke z{LPnelj*zEGu==3++^s^==f**r;2bK}pEa{Xg=)m`*DvbA3r{)cDQ z2gXijs^q@u0tHO=;0E5X2Aw6M+tRL&)~gOE+!gYfeIvoz&Ci;x9Gwf$4IBbBUX*za zELLK0`SEox=l82+)~7pRP+M@S8z8!d>CQ6~*)oW#Qu+2a2Ylzxw<|DdL$B~y95|km zw|T*YqWK;9TP~<4jMKX5LGpF}T55tZpyM+FtJJiest!|Hxi7&RHy+IzkJPIQz;?}p zbmT(p13}Dznw#M3|FTBtg>Nq%Xo7doe# z0n>IFBe{+eiXLRdqUX@ltYzqK;OYZd7pw+3$L6bP-Z~6rw(8(BLZvZP8xAsp{!pgi z@m9nX5Z#9$@Vf$3x(Qd2ODVCQk$DSgC5apAi{v9qk2j#cN8yT;@to8x)Mltaw2 z9_~9wkx8$kp6i~Rj5X82>-(MJvyrE~exKv##}~7$j4Cw)r=!tCwFaqBux!nvx>&jW z(lVPb0fTMSMMde)kLQjBc-~}>Z@c_#WZoP8MByacH>^2wl80PQ(c2sfu2XV^JeCSo zHlr}^dJ6NB%$fKY=eV_ez}_r$e0T-k`8xkn>{(cND{s9hSI=%Di)G}=Jwahdy-U-n)3*mfSD#6hwot= zW%vjINXDXgQH2{^GUg@j{o(*oz2gB>iq z>3t9JW+4hLeW~jog?fwv*u11R~NdQl>EeBl=g87zf2)b$@Qcwo^#@#xI-~cfXy62+8idcdDXB$CIL9j?6 zGur#8nh;K=Fxx3 zVYuwL;}<&uG4R2}?IAaqMH1A5NACh)xiv2BQP4r?$UO0L6(B;-M}%`?929}D{vWU8 z3RE4;^0A(vpwm+q0a>bL;%Q2NY~coBg7sLdbfI;2rEW_$*-*Y_JUu_-VsL?td2>;E ztcZl`_GH*#eU?!KNvfU4=%;}=2OZu0p@7^#xz*RC+((0D#m;dkG*a?iK;Gq#oC>cT zKpcwuyCR=L^{TY_R}X~ZSzqA$M0LeY9}oBV(xVLdGUtqVyu+S<-J+lB02!nh@tCfj zdFPRct2NK9xn(vV3;3zj5k%%@p4HY}LUvA+S`JA)V0}P#I)8ae7_a{MmFPts7KpOQ zKN5ks$P#LG}yCCP~$Hnfn6n4cz!P1>(BlX_qWfk@(zJdH%Q_FWN0Ev=2#V54bb z=T)+PML{>l&UZA6Q8i=pB???1HrQxs=mFCOc{7rUMzeCjNyghWe1BUx8qj+;WXao@ zI2JvV_@O`6E5+4fzNS}BBx0AKJNn2-7}%rY^{} z0T9>WGvScNR=UvQlc3K2Ma8eq_LD_mh4vT0-H!@oIU6gCv^?RNKC3Z>9gf$u6S3E3 zd~zYOF3PW0mr_@@ZHZMiq=O>ok9a|(|2}m5saQ<&ZYrm-_#S^eo{sLewXf!->su!#klZXUh1oCI zs^BrZ-ntd_`z=B0TJ15kmHE2U-s0{gLPVjPB5#<(lepRozRjicx0=vEf|X_KJk!{i zG>Ts(jnLPGySF(j^+O?QuHUYx)bbPSaY=tn3OWpNO$-pR6D2E6xW3(z0e_2&#$G7| zk8IxHD5BoxhpUB%F)lvqNr&(|2zQdX2a~0ucPM(CpRsHCDbqL@46*nSen_PfAl{6Z z0HXMaI6CQolo%Tki%;?usR6e?wy$$*927+OtG;A>%-Qv~(AjDm{}~!Jj$}B6pPut^ zLC=Ph(=tR)73SU7&z;DU3KTTsxV>ZSxo!m+QFY55rQZ}pW()XG0}`1vKxd~}_ZZ4i zd6!OUc93}Y=N$UNW~jQ0Ob}Cw@aSp=8W!a;SA-T}-7YK5Qj`oM6vWk!7&bxT+mSgQ z{R@4X6ka+~gTwq^7^o4x&vY8uFbj6&yKZM6un;;l7?p#jI$84T?bXS9Y#YL*u+y=K zgqc6x$GSyo(^JH<(7@K8)LWp|Z6C+XMyXWE$`1T;+}hoNlfFLua!#Na#C02)rJslc z5cqlo8B#n*FF;b%ZM%Gm_jQ59)_ztU`Ow8E+_^3!X)n(iqllX!bi_mA`BvLe+$p2Z zK((v+@wb#~%ZFe_`%Ts+)Da&Ky%&Id1cyhQH?G?etp?kpACPcea4ZM#*!sGTFJ`+d zQHcx}4oc@jKIL->iFR)ZQ6ttD$gw%G+YfQr*rDClQis?yhVouD>yPtHtdDluXKjx* za~x_1NCX?eWuGmwxm`%9n=vp4b}`^_@>7H&J+pq-FNvB-w);pt0F!3j;D9qUu<6HG zs??Rwn?Kn-qkfrTSnU`dr>M^h5!ey)M`1-JH zRHxJ$e*0qsYTdqM)uyA?ZH_tUtUTDaZZ&OLD&3uD+(8MF9ZvWqu#KaIgfksx32fD* zvEdN?vv&36QHtiZ(5r zueeuQc1kSzz|S0K+oo2=ku~Xwj5ao2IY`!(PB^$dd%lWNezJLHP0B&@CSO=XLMjGj zN8oI5J%(qBO47~@ds?s^j@uLEoXkh7m`d$F^!FItSSl%309_m2-xwu)zIlN#g88xY z6l)1N!NmdRUzZ7g7x-Xg@OjP6M-MW&cnOLB`A+cOIb%QqMeUp;L-?aYn_-Cp5sSS~ zqWapqRk_3|&!Cty+AxlnXCq`)%gne-FB0Auf2rMOL0>V=C1PCc>E0qmPlmjE=NcO} z0z0IF279+)hg2hTQ&-p_6rA&yC zB{|Flgh?YF6CATc^^whhvjjGV3%F?Uw7t#HM!j-HvZTspE7&k<6EuLyM?Ij;i}WD& zZK1L1X|ON$!07q=js~)OH}!*6y-XT6%#qZ7kviU=ho`cmMejQp#D3ZKzzj{F0?Bqe zs0Ak%8zBkRE-53}=dR5(TBis~etcVSoI|^aYO{3o`P4Jfz-~>Sat3vJ(ODHX2b!QG z&j?&Tq}wEjbjzNXFsM1-VG;jn1{Al2yoPh+D&%hyE88KekNluG)=%39Wv>$=JGTyG zYEQREt3UGorBqUV=7qbzmd`gjR)TvzEnq&z$LIO$`ei^A=_@q)0wb^XO*i?Z?{@31 z9!G&UE|0Fo^I{hfWxPeqxT87-@FQxN7cUou(VQ2d(g6Pf>MmCYL4h}o~p4F(1R{)5N?YE%UMJ+Isoft&9l{9E>|w`Y;v5gB7CxDzsQS$iB&YRfUJaikq1v!qy!tIwd$cUxUDhAkHxl9A{EXk)*tsBD-!4ysu*{2L zt!W_uI4nnSn+T!S^IDwbVkH^>X`wE(3`fTMvUO^ps!5$0eXgBWN<^4W;6?7Y?eVjm z?;pTRlul{KJ%z`g@orNfGS$t=9eSMOzD*1Uj*EzVHY1+u?~F;42E2Tmnx+_5Ji{L4 zD*zivF^Upyg5Q(>SD%5AMgk&E@6*JX(%#?X_Aros5uf={<>fhENuhVfpLUgmY?bag zLQ%UGF3<7YDZQm`XS!6RabVKxkEUYW?E*RWx^!utcgk=x(?}8gBIjIH*72@mjW>eyesXSgZ+pw+h&btD}Tn zRD76$T~y@zz%DA@SNlup=dTeA>!3VOt;{#!N3lT-dI_!W~v9_5$IGKM)`; zLp+vtoa>qXmn*zQ9r4i31s7Vp1-h^B_ASLQ07w++6WEgXgdF~FMh)Gn;LueH-&R^@ zm(5geqb;PTh?X7G$z7tw9ZhZB<9xmV_7aoxD0rRyOVNt#UC0g+QCmjS6{=;hFQC`8 z*t}RY)|CMXSc_rFnOGViGvbjh@`FNzi)_lU0D8ZSU3DkC-e0B$yLA%Xhuu0&b--?& z8WzL~*foxo6o508)KnT_SWP0LXS<+}xzCHXqwL!JVS2X}pD#QZFG3psFFy9cn4ZP` z7o@QdM-49p=|8>ncZY)Bp8O*#BIL*ywTPBD)45lJ@PEw9d11Vi309BL?;+yZuNq;S z()#~nM3d2vXsg>ohqYmJ|6(lZx|i(#^weFSX|IAVrpeFT)r&R}T;$1iSg>^g(^xQ- zD$oVr(`F~uSJuAOkmuYu_VT^cCMYKEO9Li#(KS1SE&%nX)U3%P^+x|cEGiE|hw|b- zO?R&wUXynY$fV(4!v2$p%+S? z;UA`A^?rP}>VJ*`TP(Nfdm)1wI|+%jy}~qrwTnjQA|n?|wUM{L#mz*eQFtCS?tMn4 z_{QVds8U|{l0u4Spdy~BSo{{rPK};(6o^{LIu>viciMHJ-A2S`Ouop7Z1^qEtCdZr z78Zc;&61kDiQ_L@W@xu^Z>yEO$hMvnh`>T#%5_gmj_`TjeS9Hcl$=LX&)d7@qgR>} zSA>y3$6G(E&-|K9*9XfOV;)OcKiP&F=TKEo{gZCrEdf6*v0J*Eg*)A&PRk2=X?)WE zqep6swZeUl>&NDvlX57MCb>k|*HI=}!rIF?7@ALHzYX5T->`mW8uA_JC<#JyLkVXL!PPDQ2Kihn*^m zBsNWD*GYuZ3ouRgS-(!m@>lrVW1YzdVR1*M#Z`9kH@R0=Zkb6lW(L1;)ywS?bteWjI3cUHfOktaYzg>s8pWLol6dqh-d}{$Zz>YuJ_SIW=>%p^F zaSV*ZW?ZE8-X1`w)^dIYyWUrz85#Ltr|l>_LbGp|xm*#-s<{R(ie1=!y2o$ zaurh9Rx*io2o6{V9;KVuEqZ!5b~Vb#0*8EzX{Q*9Wq^-$den=Xtm*;5aDyYx2#kx_ zpyN#}nrR^_xjd2+tfsw?FUlsdrEptfin+!5Kw8bj$}QSW+on`~irv{fBgHBTtR`Yu z)P%>Jac*okn@Fg)Y<-q&=VftSG4dJkEz*512w%Oqx4L=a3I=*wm1-n*1=XGVFH zX)prKS2A^d%l8L`YpIC464VVQv&G%KrorKD#>*PPn6NzC- zRi3!d>?+(+ia5V_dr7;TP(xx9yzvz-Z*b|`x9`k=%%5;nPSvLb;i8YycD$*NTaoP4 zVH{M-z_BH)la-Y>X+^x~dcdKZgahUgT-NuXv+TQ@zAXc6fzD}*@=1SZ0!9`m2;hs6 ztnG=+L*`FpxMJ@!R3{so*hz!1ZC`jy#Dx~7uE#cNEJa%Cm!Kr$rb8tq_qe!TZJ^to zH~Z`3H=mi$N3;DpEM3R*BCcalnHsGrq{_(@kLr zqU#}EGNkORTU3Mj`sB(xjkNbEg|;(UIk~S)U~c^}0N^SY_>(#_Zk?|4=^s=dKO!e5 z{mHPKypB4~446M^f!A-#Q7M#Iv)w?(J1{7cvO`BX5Mc> z*ZX$`8YdHz+dP34e=_vL{_&c1#7`g3bA1VuH3$E6gLl&S8VMiGvF*xVWfbvG&d#yo z=a5|dc+wlg&KEkuN65H{Jt^iv2_y1-c&{=_`iE%rZm|>L)5klicH(5$9+)Uj-iK{0 zTF%_DTwi=NBqwkH6XW*t`R9L57YMKjIr6G<orhROcscyf}!ytTDCk^xbyccVenhO|z9Gf9>fqaxuV zuJH8w(Wc_SJay!+nz8gL>cDmHc4aI4!4pkZ(A6AC7*6DTmx zaQy*aD+b3@aLZR3`42pQtf?t5uTcori-c!skQ@NNmz;}D<{zBI=;cOf*6r_+!ave^ zLCqJ($@+iP3%Y`@>q;h=*G8KE;GX}%$v@#VYmM^2ClMOV!!=JR7BwJ8&Kti(rzc#J z3Q&H#*tPxgVO>zVJR0+cgL8k_%Y6x*T5zNmG?tSM^d%7>e*2Q0}SPxq~~ z|8ElN(=Dt82Vzn9XWQui@&2E8VE^OX{DgXg-3ofGn$TcMJjje7uY$K_PaMuY+Kn@d z_?j%-G=|4Eau1`}g!dH;E+@QGIWJ2eJ5w0`-0K(7IL2$T!71*eO#Pu0v2QWMVF1{f zv5<$-!99zm5`~*5%CABV6X)pL%S;ki{uoosRgeq4R#0|hltr6G+Ub;33EHgD?s^Tk zs(ga_d^t@o#hvUc>4|Xes*Ls*EFNAA&DPeG`|rQL#VjJ0IvQ1_wnSr6vRS$!j{l+m zLT!Xs_G)oJ2VtSK;4|BvLptJ_E3ld3k&Ty zfa+6Y!L@Xg%4OT=`gsqdi}^C|ik`>NK^Asu4ykRVZJdA6XBw+MV%-B{e(|RA@jTh3 zn>aYRGq15$8y1pHt?ZtPwmZd)|`Eh{AC+YUKehrCbU}&i&+v7;GLd z(Mg390~F44gyn?~<_Ha9mwDf){|eh1dK*upx;drf2mxXUhXoeWsQ*YB@jlugN>p7B zwbfdFFd})}2hf()`OYUIshcMAvF9K=_w!0VIn~^1s;3|?a=HJ~Mnu*lLlqop2m@p) zNwmGvXn6r4z(bj|lu5UkrX*U%;@2+sx9i=vS{}NL+F7nl;d2!8rN*39_9)_MmlZ{% zv*-J+KY8AI@nIBwZfig5*{-0xmA@_^E{G>BTp?-FDVDWX*EQb)dv>(tdp!$sJW>L- z>8x&<>l#%hC1x2Vu#D;vm0Nv9mSS5iMAGd1m4kW!WD#bQCg9;$(;&ckJjDCeusL6m zOQoTO>42B(RnS%*9omv;O1v+9N4fOM-JUZ|XBDVo>O|qD^(cDa?zey3^DCgu&M}Y7 z3P8RrM6@y*5a;=8KHRW)FEUUIw_s%=?1-eyzJa8Q6H$N4n)7+y$8Q)M8&)0u38lzZ z%*fee4bHl9m?;MqehokGU*&cFj1FiteITuHN^|0~@5^A!**w7ke# zJ(g^4U&o*mw@$Bm^({WftrS6l6lYGj*H(EhnQY~yz7QGZgF!LsQ}_t=9C0~P(zo7M zbX*(|urC|}GMbBt7?D=(#?Mu>E?eqiPD3xlSZv86CXp$(fM~1UQw4;43J;Gp zL?Ou^ll+yj^a|pd4H|McT+Y!`ZuAmN${;m5`_|DWn<(^YK1ZCk9yAa<+8R(qXN1M5 z0>E_BThX+ZcQj=lWrnFry>1RtU%4ou%X088`&Vn!9+Fl4E^?^FmX#D`mBQGHJ()1 zmYVvdDE}p}i{~ewLE>1q>5WXI=RB)1>C(HI8%G%jH-cFU(h;4~E*&F*OzYLiY@}+b z+o)oyQr5|yCL;e!mH!ATp`MD+4mLlf8rKs$c<($+uzcw6_cq&8oh3!qx_pdFfV#fZ z`+BPxfi^76`sZMePzvoTq29MP+FNU6u8V0^KH2N(YDY9DT_yOF$W7{rmg0acR3$eouMc>;w4o(E#@Y+T?4+TvBwO!+@+iRk& z-4VT!0{2CFrpx$7OX%jaOhCi~{*7^M-kqmnvLB%2>z6mOMIV#VI}pw#Lfh-Xi;Mb9 zzksUx0x{>aDJJq>+q%BBD(V`JBZ}juAhABDjh|E3>9X>&3!dL7LU=pr_-P%E2OD0& zyc{PlVR^0qJSvW)RNjWrGiJS>{5B(AasXMHa9kn$;_zjpf0N7N7mpKn74wh#%X12h zy-uv}tQQEqFbT0@&&><(h1%>a6S2HkwU1krPR#tB2h=Gg<@}|l@iw1lOKoBXQ82MxyP-?{7cL-VQvE%iolWrf1-@LRv<-Dbv9t7S+wf9pufON8l11^J`c zp=iIT$*yT?>PD+uLWKYlc7+!lyUzQ|EfV|=^kF}$IQMpdkKYjkl4%3oTr5Kwi&j&= z*;Ax)KKd*+<`A5`hI?F?`UbcCT)d*$r!Y`P!W3nG!~}s2NZpclQT0QFB6+jEab0QWavvzn z&;lL42@bF1UtO739mxb`2z};BNUd^vMZ9VKY$dQ!u-J=de5^(>t+xIBk1E4W-z^E6 zPaL;dNu4*NI@%5m!UxY#^5!LzbL2+6AaYyES(yX3857dy$Z4w|NoiN={$@md&kv(K ztas^?`_zU$&6j$GvCRbsoV9Jfnukg{-<1~o9GW-9D*-t1- zBz1xq_{eWu7<5=aM2g^O`7!#25dP+urXyHPkL6@v&9CP%hP@k7c|HbkahwjiuKi)J(m z`BNmOdE@lzGfr(?UIiH=e1qqI<}a^)`#C;m%ck63miWO_wv=nuj(2d`W5Mb8DEstk&X$Q0c6&{`f-F|YE!|Lk z*#E@y6MI%PJc|f> zbDNNcsnCy`&@frXr-gzlP3h(T)Bk=*aYwrIi>>`8%fUQ~0@>8ByQeZTt7EmZarDmm zXcOn+8&k=BhD{39lhcf5pyU{e#a*ndr1OSaEd0x@oePL%UXz)8KI=fm2wP3fPQikV z5JHZk_8nGq{w5PoQg@%M zIMcJi+rexA+xD>}_C2S>TW^`+qnq zx!@OI+q&0xBbb$+)+Nj?)Gd|Lo$@77WBR{*{6JAvt|NKq0!Rrc=jHmQ$knsVUJTNV zftNUVajt+m2U@{r)0qEf!7yXR33^3l)s}K)KXq{;)@g(6NoaZZQPfs`z9j8GOK7|I zi}?Yd#Z>~fBtWMLZh^OqGoi2rm4XJ_ShCDQjahK@u=DYgRS;#Sx2+PsvipQ!zgY)7u znUK(CjK6Ut1tEFPao(`ORYV|PvSOBCKpr|1I3D)2u%k|(g?h)}J(9gKE;*vLO*i+& z#{|{GMqcA?6fhY~dRM&LGmd-i3981EQo;R`DQT)6i|1lD6;GN|DTE*S zLdj2H|E%s$VRuh1ip01J@J1I!7P?S$dTpQ^LSwty`?`Xv8~ccdp-fj4Zm3qpX7|F4%T z4}@xMjn$_B84snVRh$ucAGFxFl z2@{r(kfg+B!V!6tPwy!;mr@g@j}4Icn@!!7S?SV4^aKwCv7nE;6Hq-EE?~jOr6sM( za@IlcLf^x==(}!#cGc{vJB8wnZ6YVAohGjp#-MdZ9+qHZb~Otq8J=mILFRtEU0|2~ zO#hNa6-~^q>5yZE!vIF)16U4UT}S8~{q@r0>evN2+({D*`+UJFN2vc>`??wX$Zr*Q z2%JTGo0U5P(>y_-Q7gJ!qUjsSbY>6J4CspwO8GV<-yn8H=$j5hZBvt{a0eh{j}Dbh zf6U?yN^lt5IERmCGrhBkvl89ewEk*GM`xM#8T1&(R0u}3JpCAH_j zoGDdsgk2wDdHB`XQCFOGY4o5m;c9T2K74Y*`lN|)Znn_v679FQFpzG|3Y6C87Mm+c zdRUX!vQjrHc)t~V0IH_XFsYJY)`&(#Q_j^-*_RB5zMz&oYCY?bY7A$x$EETb@0*rG zs6nGX39A;K7S&8;FY>p@k59VaAq-Ow)Qy4EOcWS#i&`62rUWo``hAgYvmy+vKbTgP zv?@wF(!%-e{w-kZu~$CcF~_ghO`*A#yfS`geImS5k^`^42SiR5mS#N;zzEuTbTiRk zaj^_LW)i=;|K&$eIf;7j-E=SH6%AElN zK_WDfl8NnidpFz7e+lwRGJ7WuyUA_0yEBtG&j|P-~)bLfNSc4Gy9nb^-^8JSs2y}dOK-5R-N>)Q;R-anC@MQ!CiVK0L1h@a z0D{8YXR@nX2Xv*=JTZaF&q(SiQGsnVk6L1H!K}!i6Q}G4AQU(7$A(|~ptJagNs4U^ z^SFpDy2s{pjy+R3YXbObHE6GtcgrN6foll(`6SNfdNon4mYmBbZsU=oF+R zZWSjxlpHxyMx3vU8`OQ{&LkVGO6~Jr6Kj$x+Lz?M%XAZ<5wCxhBa*0z!?sa(^F_&W zK{Sqh|4$REU&U`3HK;Fgb(z~;HCh-az!7x?k78NInt1o*D*H*^}$yELySH=#Y1~9G#eiu zXE)h+v(V926DS#>QS)v4fvT{$XAx2;FlJgzw<+1fDi?H=?>vXuC zH@&i8k13=W{(hM5O2)<>g+KYgLiZFqwN6VbG7Jud*WQjnNshju);(LLm3lK+T7_<$ zbw*;7*L+aO$?FUu#LI1^f*#l+oMyz`**5F+$wUDT6q8(ud`HlRM4za@HwnYYvHG{d zqCCu?!<=`y1?%8vSNf%1Plq!>MoS(+d)L0>;uMK{^?%0c)WO%l4M(0(m+gWi;jr44BHXBIjk8d1Bjw4c!*_|{_df>A$Gbe7{B6{ca#iU_gp|u+z(vgv?4t5y#1zO#@G0eB_vq#>O2G@Y?Xy zWM08m)gIf(uD}FVT^usaL;q2S8Wn4F*Fr4()}8rm#GU;ib_7Y6I_z^o1#^YEKr&9LK%m*d zuYoz`(Qz$5GUZ9)ItHl#pM||N_sdM8ju9M;e9^j3g7G}p!gDahWsdr4YcUq*>!GA-yH-Xp1gDbY4{AfYAY z0)~I25M_V*QukLP{bXTXsQY4bbUEeA4VNa39(}kUk4-M47dFKCo(6Sp;s_jDOgMX| zZ$FQqK-ET2XB?=B{jR;FnZFS%`2ud%=Zx;8n$v2McEC+fQChoRn14kC2{g73bKawt z5(+2LzUf*9>S$(OYy$rD0&h{s#9*&M0q(2R+SF|}0~xnmgSxrJnI_sK=vAK>ufw%gp#7xv z-kauFfXDNywbP~~C(tco!z&{0kb72ME0e56WD9kz@|losGcWhz>|igxAp~{u5N?nN zl2sbkG?=zN$S{NkOB+E}9=I5IcGvO~)XH=buRAME6Y;JSy)}66Y5eNB)$sP~tSF&*EP+e7#yNljfN2109H`DAq+#`@)4A_dVg9--c92oiN8 zNne^s`&%{G#8V1;7`=Qlz>w z%16P;t~$Gds1*e~Qzdt% z%x$QI>2ae!xwA#!yIr31z~*{K-tGJQsBjI>yLalq%L7L)SG2HgH~d>z%>(YohgNs!; z{!g%k1`R#nHu^KsTTxbvJOF~3Z0WWGCG7w^YO`faGZ7!n7mj14;vje!C6HGfTK!Y) z+DB$B-gO@ODa0iIX}y9=!T$v^+l+$l79gIV-b>dL!?-ozl?j3Ahra-#Q;tT;5~Clcp00ty_Y7YCRtDCi0C77cmd(J?MO z3hbW5k+ARIwkH+!MA#-T>Xpi0&>hX+?bwG;KaB)V;bELWsD*H0DWkcB?ySf|XW9RP z0%Ya(ETw3J>1XPUNtYLhAUqXXMKxo9?Y;MLTQ!N&w*@RVJRvb^5(S5#i%K5+i2w`{3TE5#hSvgYD_h8Fl`q%;lDO?Uy|x`fuX#X=JW8l<_m8Ty;7-Y^ zqu&A|`A=r}ZRR<&%n@n2asR$c@z}k$D06)c`I{)AV8a}~u0v($w0rx@p&EUK$|ZH1 zykI^4-K&?}dRN8B802r}V`7QM!ww`+O~Zj*bGyoxo;fA17x)yC20&Cow_U=NYl-VP zMnKj#pA1WhyNm=DV73h{D zC*x2|$M{YGW8sKA_Ib=D2$0o%!(1YSc30Y!J>Lo6T_ve# z&0C9Yv0IQm?F?Z1c>>-g`9F)I9y$EuJYo+z>% literal 0 HcmV?d00001 diff --git a/examples/img/forecasting_pipeline.png b/examples/img/forecasting_pipeline.png new file mode 100644 index 0000000000000000000000000000000000000000..6405ed2482539a47216cd16584aef551963f2bb5 GIT binary patch literal 24834 zcmdRVbyQqS^B_skpuwHs!7YTqAVGoz4-nkleXu}+h2Rds-Q8u7;1Jy10zrqtAxN0L zB=5cN{l49E_UxZ~57Ts4S65e8S65e8-y5u?Acgsu^f3Yg0;Y_##9IUegQ$`2^k*MnHi7jRU~{RQ*Z+RQc7c{wd-hARL@s z>;H|D-MbASWc)6E1aiOWBnqnmnLqefh)7?DX8sh>a0s`t@VV4nf#TK8FQ7*k6{qgS zBLLG{!_xwQ0%JzU2GWX=Js{QjpIiYDD+e$l%DseuwyVIuzk50O-HMHqdy#;`LhC&M z*Vhl|`(1nb^A;!q_)lLd{4=7J{YwBC+q&@wQh|U__jTYN4G#YU0r3AkqvVXQ?2{>< z(%dtIfG|A&L&z!y$N&g`FmL|O01*G*IM`(L|J2GUsRFP+m^kE&zv~gAle7O#;ZN-} z2@L=oUOO40WH!?C1%R5ypRsWOdH*$1Dci(9m_S<1{tAEt571F@Z7YyL#li>R&C`I& z{MDU(!NMnL6QJ1hHv~Xf+`RxG!lL(jqMb4gr1DlVDF_G>_Q^mVK%f&*4+x6|^7*_n zKru}~$pHaj`{;^E(hvXwkG7#100;U7oOI@gK&5w+`*HOHTMrNdf#7P+tLu%fY2z!6D!VI1OA?2@Y?E z>u4w;_5pQo4B&7K#K}>(jfDXfJ{)5)0X`e#59e@zljCo!M!tcE0!Z)SaQ8bp`2Kbx zJockIoE3fwKg>vh!m!C2;BcoJ1hJ0bT=_)0Sw5z|9NsRD0jN zy;nRm2Wd@b1O%cd_rC|<7*0hH5Okl&NQkPs>m9U48k28(Y@-aiy7oVK!O};h%~_hw zSgc+j@n%&QBeAVHpA59gU#E7mSFeOi+}-s3Qvz8+7(TDU@|S9YH}xKZS@OHhGt-~r zlwO1^_!Yih92jUNH0LZSu6oruEi$>?=VoT^W`6$%JndAtpu=0wZI3PQj7uPwN}a2- zgy)N9`V=*aIBH~iM(o?>%FqxL%-PUPintgQg9TRaIh&r=9FjA3m7EdN`si%xd{{nL z_-nq|dm^sXGyw|DQsKSy zLqIFIuDT=;d9~X8-Tx8N!9&6jUy=cXq?cb}<>2yzRf)^v=){MKUJhsJ(lK6N;!zWV zAF1Gxd~IcjY|iK?;7lLsZkpZHygc>w;!2lzcWLES=jmNkueMk?P56K4!hx;q znX_?+osyB&w0~~O7LvE1;GLtldJ^0Y-Wp1hO=P#Vvr89n-MhUyolPEObUIkLQYm=B z$C1YG3>99K9&_PnBwu?Y`L+pX**{>&UCzCy)Z7 zhulq!@*x;l@8}HJ2(T*U$ld6YqMNr9`9C?>4=&Kg8P6|u@usrET#nP z52A**OBxNqsNO0508W%C# zD7=$oGq<#0Y26?eSY|G0%+)K31gExSrG1ofsED^=Gf&F&DjRFX_4Tg-bOV@Yj0EZc zI=@%SoOy`-51zd3)Wx`YItCHZwx~}qwLhHP!Lt?018&HR^>lA20_FX?`fi;Esg zpx0v?^YO$_wSulUOf7oT@h;!>+!8u)*loy!4c^YbdUIrT3VWw2hothR>%}Lb&#APz zQx=(@q6@^k)4t=$`qn;~skPAlE^;&Bq$RfKvt(U@{!Ex*sMw2Ws1@Q133{~FJz2Ox z|D`T}VmT$r`4YCb+`W?Qm+uy%*Xq+)W_R$-q}9Kn8`fmp`dOe90S&I-3J(l2xxEWcl zR6JVM(d*4MTDCr6y5w=k-{7iF)UzF6cbuI)P#;*+BkecBVQ=3J!K)8{;tDr~3ttXS$6s#h(PDjJrOAMbE%Ec8WbbMPD{(=hP zj1|G?Z~EOpfm?$C+04a|4(8gjYyLj$YwO8NK0uR2&X5L0M_|7!lR^~ww^q=uWCeT%o%(+#^dUGz*i8Llb)1&4Jm34?i7IiZ1EMBkpW55>#>zAnm2KF$F#6!> zzcaC(Ge`_aWo>Dz^0wrNnt_Ph8r-ObSh;P5tZ+7$h;SRbnyuQ=9z;sgf^U2R?1tej zM={(c>*w;EHA2oK=p0{C_}$#LQiHw_tQ5C(zcaJF85L{igP~VR;Ta+F1a%BGp>xOG z`X>cGuzDmzE8WcT#Vaw2%U|6OTi$(blC97&Ig?gMpENLYjh4Ib@Xfqs=r7jvkJ?kCP9sB3R%Hd|jb36&wgIz-_@Gq2Zh5B>=QNl=&V zu}}T7FdmO=_tgKU*43MDv2OWISP&;`zt_S;yqz{k^7_RGU2An(#k*W?13&hsl}Fb| zUNG2@UtAw_sA)QhrNcEX8-A{A>MdrwC}yM|@xpxbm7@ZKtYdmg6K!9{Bf;U)p*K;M z7jAT3Xh+Thw(1!kDe>}ab9k7hEH8p^#H3dBbqm`!X=p+MK2D;^Ju~xG%$#K;>uo`FhXvOyE78owHi~1f*auKQ0nk=OL%L_{z2e2 zsi{4y)kMZ6MPNg1j>kp5+l42Oruk-IpqXi&NaVb|Mih0du6uaz?D@grYs++Dh;3bF ze#Oi}vF+|)+8$q>osFeoIaInSgnT4OEq4^EO!$bWDe!Vba>Ff}wfV@VGRt1N<}l>C z#OJ-_%dN_zn4cblkH`WE(i6gn!m)tCS;PvFThDxO&~T^t{EZqZgt^5R5JJ{y0+LND zipERyeridn0W?~UmZ7KqE}vKgb5ZgSJG;tBztm$2&Rg6P#B?V`b$Yqw9H&Ufe3p6E zau~Pc?UI*$7MMuq+|5*ew3(@;sC?z?D$##(q~WO5U62i^Cg+491gOgbhs_EUs94ja*u2 z7^&M5am+pYR!=1yr|Q(4bifibP@k|ISn|tU>TUF|FPUbVm#{+v>5w%gEZL~Aj{c_s zINZ7cRN*8E(xPnRNi3d7_f*F(Jz@~K%NMo4T$~XTRz2p?FPq^NE78kl+qviF`gj*I zRKFS{175G`<4l;ABTn#ZW16}0?K4A&Y!q;3am_8-CNu@l>~ZHKu?}KizhI_YuI{Op z{o1xhM_pNqA zmM=_|(ChAnR`Eq&o|2$N^lI&M?(mG-YDzjto*Y+;k(UM)cCQm_1AHF^ zr#$u5+qojmJ40DRv5tYJvyAV+9qW66B|L84+w+kLX#sr~L**IBry^JJR6?<)Og9U$ zZ0TR?a%1pHYhB8A`kcm?KbC5My=nIH%~kx8(AO_#%+8!$w{xd!6|IJcbd%DW*iT&& zd`3s^2HIvKOLw)BW#BrPP@7#pZ!}hoI=0cFHyvrQBnxKfF9j@nwxVE&ZWU+jp$}wB zP&{`lG2`B`e31fl4lti>YwkSg{k~!7apHtgk4T8n`0mb82JN7gPzIue<;~9O6det4 zEc&3Gli|EPJ1b-$(3Gz!H%_;*9{gkgQ3Gtu$Oc;wL7sjHw8@QM4xi$~u7UQ-Yt1hk z&BrI7z2H$hJQGhh3CpD_$=yp^X(@Sc`t^{08>3v>^Z|da-|L!>6TdrQvX!G?a0<7u7)P3dLaJT&h_3CFt2W(ZX8Vu?Lo;~g;t56Ftk@&pM;)1 z6WQPyrRq9U?x-gFA}n{ar}FI8#)1^A>+HSZJf_@-e184KTh7O0Uxv@uKAG}y`p4t6plTa5Xn=e0(l1fB!*j$)32HH8M(@W{>P9c-YhUEo zL6gb7?s$MP+#<=jQO38|dqXrpzN%6zGm^d4prZLV)=f0K3wqwl3&RWYO!AoBFLO8` z1M4+qD14DS9pP<5u6pvzkoG}BoQ@3=G<_r-58)K4-kpQ*p(0#e%|dnWQ@ zy575_>fK`0DyHRZuBo;#KL?7x?CG`I{LIUi?b(rMiDB{^Nh-Y_8w+*n{%gKa)64DI z@>Z4V=vzAT`R>A z+tE37YM&to)yO-CXZA6wB~&}d24Q!?0=KeBgIFYInRQT;&c(#p89n`2BTu`7sw1(H zNo#v`I#}`!*%#17nZ%r@_I%oyJ(8?cY2ZruLeW}VbbXTOk-PblrNeTJpQrM-@o%tt z1K%k`I>UXCX)np%i+}H^#Y*}UJt*0$Y9_jY0f+wE8`AU(9 zt2*GNr6tSMx^{9*2ymuX zPtHxo^<)lJiO9KVfP-%>bB?*_(Ahwim?@6tDmH~b^2fm{QZ^x-)w0>+KhbmJI=_`O z=tcrWCoB9@1e6CA7^Fo491}0tBR(vB0W?QR2)R z&w(wOfHgS~jTxfo_99VbBsXpu*eTon=pNv`1*7#nGokqgI}Pvkil&oi$b#awJIFF* zLiG%t=L4$~mm|r-SL@zOqtJ7JD06a^bmn(uz(Rv8h#I!fK&@ z!y5Cp8UwqP71^=X*3Vab!=!PUOx?W7+Ajrjt4TOOt?ko;ynQ9kV=O!24SxX$jdUOL~EcfrJ7*x$9rcn@4%kw=IY48?=;65O*^Yg zb}_dEmNYJ^K}{{H`m@`pfc=j3ETsvVjrSe2Q_=vpO3y;@?+qp|Y{FnZoZ|mzU6C%Y z0n_)#Er?<_;QOcW+ef%hVo|Q^y+0D5Qj7a%?ma4fg-Sw-_DrWpCeTsp+Sty{?nP7v zf0&y&`zxa@g}A*pl_A}YXyv8lb|@XxlHvJ`-d&Bf9yQj}M|(OrOXCupHO0y`_M@r& zKLZ2TLKNPW?>*$1b#>ybf#B#Qc2B#%nbV81}vo5lg-PO@vJ!AJTKkEu2 z4bPyR2k`m6;c~Ck?(4J@a>Jxo+#7ZHNTLE}Rpb4TNa~8j%$IhJlEvn}xqB@Q4l+xVHe~jyXo{rRSmCm9GDk7~jmSt? zPa2MC+yg`nZvDeUT|7z+XPQ+Dg3d-B#?DG8nqM(?u``*$k*>XoO2gdst&0#zjxL9gxhv#_exlVB8j$@hkWZ{*P4rtlubHjbF8~U%|VC& zJMP(ocTef!w@jOi2HKXbx1)(a#x=_3+qNO%>v6_D-iSDpQH7sBZqj@C*$zP)6U!Xy zJbf6w7`3KZ2S*2$=*#pzMtxj)X_0|Kx44vx&V(cBGDtdkS@`tjkD=;tR+01}w zWB&D$IkX53vTXyZ@nquS_L^mvw6f~GGih@6jEvAT%!qFN+DJ>AR|IksEPk)UT)bRt z^x$~|v&H8*>Lf|6#wj~i*p}bH4oy>ETAmB9tfoOj&$6GQwB!a$APQxl6H_TNI<9L> z{A=AIo5&JpPw|hrTMy5b$JZ2a-wwvLfriMOUj~Y>Ux%&)^u25 z0pgH#ed!R;uXC2;;YPgjYh#>WIi9nMLxF_}NF_n?u~RmYEI0A(X6BRjwuVd5Pd4OR zk%J@>Ddj>C%h(G~29Ccxjem!G!+<>87)$!zjkm?yM(yxxO!v}W{gg$H5X?_gQ>z#3 zj}-lMc*}~@&D2fZX&h34^1V#d$nITOtkPSS+ZXm<*ghdvTZVHl-`P~QHdnOi5k2}9 zR!!qsDYV9Az+FQ_XG%ZAL=U)8f6sc< z?!#HU$H-)iCU2sjj;UW0plw9R355YF667c)eas0%Wuok*pzZ+{VXix|TL3=xMnu$i z!=)#01@DmeJ(Ai;aE4@}<<152JY=EPOMqawxv(>U0o&nC|fq z{oRv!YWG+NL%Mj;cOiTt=KFoGG{3X^-cT*S-{|_>qR@6SAj@Ic5iA8pWmYvj(?s@G zZn@&uj`Y+dwz~WBA|*Of@45pd64bxH7RX`OcE!f}Qn?>XG>|!r`dg;|_uh z$h)aqF57zirOTSg+bQ9`lztTv&r!w7wa?KMevj3NJrYC}PkCV%R&}1gBy3yTXu7W7 z*;cjDemFNJ0cn*`lyShMt*oU8XAEt7W}#vuGdX&MpOQuLhZ7PGTlcR{D`-vx+^@>a zeYpHWL#xG=vL4fpa_?BS0W0fb(_ZFo>xaF$cvKX~s;1tmE`@upa90(x6W^(x^1Pb! zo-=Ta`}@mZ1#P~7U>2;g)0&86%_ROaTKjO&c(VsZQQ8!fUtb!{o*TE&_texfsa94D zUj4gXg!kz>ZRcBCMEb8bT~9`+bgrKfLQ&?a3zY(+%}LIlxAVd_5hf@0d86y;WjvRr z6Zp2&9=p+1Wd&t0ZQ2NRW^tZ8&lqZP*^Si~U2{EI=Xrqb;m?uW&W;3+ync13SyKAw>3i86WDB2^ZE)Tf*X zD4)7LK;B^@CL`6?9>n$a{JQ((hhsrs#*BWo-h_;9=fg_Ew0h~ybdjsrQ+b7+xv=AC zS+0@pIU$By)%}CP%Lv)SsQ}}?rF3+&^_yj@%Pt|Wwd4~t^eTz+UvH@#-3D`>HPV0J zc=Nqy?K6#vYs_hO*kC_qMrR`nm*V^7Y?!#W>Qa1iUlNkmY-3G;qQ2JDqmlAW<+B34 z>5W=6;gKo-&9(?0ds>(sEsR~t{w3OoQ6V}FDixlL$x?j$Wt)-cti`QnWC$c1H$OVx zi7%tvt-0{*rXGqFT^l;;LFgTH6>3&RMYbO=z-4XR*OikUohvwnP zUIhm}qxt?5tib4kt6c213eigxV#^6-M=8Y-UZN(yXE5OIhaM*QCC+_`X~jHVIb)o> zm+0|-hS1T#yG7a_Qbj1h4BHuK+M-c6aBOI_x#Vx?r7(KdY;QBJ0AXeS7RCvK#^gTE z&`mCtr3?ebM7P&x-f;IW>=riWFm5M@>V(q#!v)J~ZRQO^FYdFp@7ypNI*vcWlG&ml zb*Ac*CJL!&41=H!psy z_gH+GYSPSEqFUS$Us8nfy_|9RaRIUH>uRz`b3c!swi|WS2ccXMU~I6X3km(yMQ^4u zE+vXtCjQ!6KmiT+Gh8yIUiPQ58frZe4SN!Uu)+R_iE@q>d9?$TTGwF7G&2N#@n(2h>79tIRI{2q_Fyz#~oceTN-G$@P?upjQ{!h|5f zIX&%lJ=F(!KSgp<_ox$~VUe7}lI?XS4-7gVOYtE>qU!i4t5Q%2dx8pR9v|jNv{boT z+rZ#TU(q-@TPXu|4F-L%lf8QZWdmK)ZlDN^HSAB-Lrq1Q{YW#s>hNN;CZ?f6VlQOo zCO%H{h3uz(_PSFUOc$+4A#%-LW9@ofabR`G|g1UPVIf zHtV^w(p{&s;9cpeZ#5A|kT_aPj;$!;(rKipkkKn?X3Y&4e{jQ2^b;9VpWC~qE582}F;q=E?oe zuoO)V%*{tfsZYRR@BvVZ3(N_zn~C)Qxw1&i7=dgQ$wZTL?&W=;=x-0_cP{F!Z)kIC z$4Nj`mZ;J>@*tkrCuGrSGFHXkhmj?ZROGIRD00KOX!0hb3M9f2LopFVbm_PZO-EZp zDq8Wcm9+d$=5}`M8c`YIaytsZ8AZ+kDY$$HNjfa$sfKPV8KN_>YjXM-W>>_foOT|x z_i22cgkeFux|Xf^T=emVVM8;{m(AgowD|=8erHd2!lh_5=Z;cx-<5G?M3tvfw$*b4 z`pSgOWJ`5%@bpt`EuFNadF*;O*NLAW2har?*_(Z`b_W9IACwx;MWjDKM>g zo*#`%G3lYCVUYh&rdtpfdblLqH>ffNZNnNej5h>ByB{_ylr_CGE>df!R&8d>h_E0b zMt}4fAvlMB!*cb(0nbrqV|Bd*XBRQpF!!k@zr^w8<*=%M8HP+&pJH$b6-T&4+b7$o(AK%h_gaJc#q59$&f7C&J1>RjXdP= zTB2@gK9SO3$Lx*=a~i?whR{7R(MBALHu@&sw;62I#&V4xq;X z>C@1DE@wZ(bIEc9H&^{^#m+tZbGZ-D z4kz#EY9UhpfjdM+&VtdQxk1YV3Y-O&>hl9?7jFqvj`F(ggNI&Il2VTb*E!#7=rSxw znn--O+12ieR=dU$x#>!Z1Nx}mSqX|0C{)bR z9aoqoC`;CpcYWe*pZ{+BmC6NHo2~cl`*$C*D3%6H>GAL z&B?Kz6nh&}kL$0cO=_&xSAtaCjLb?5)dlnFLnmP+ta}M z7HIUuX6z@7S)4XojGJ^6tMNSsq4NS7v~3AJ8hLwN{_Uc2U+`z>dV*_S#@Ew>dGPb} z?`9Bs+?cImAyhIQd*S`fQ`&phTNs$%#-GoQa9Rl#;#>3&BbE5cS2aJg&J@rN$Lvz* z%|6Yd1ShNTZA>@yKX_v<3sIFZsL|bF z+(Fz83ki@N%zDucP&^jd5MVG?!1zQl5dRDc_TSgy6>_St<8krsI4im8d@?nyw#yTY zd)=6p@XGrF79E8TzK)3>shr%A%+`PY>Ka4fbtKvAVpkr=8ru@XZ@kK%!;Y_>V1Qy+ zl+P0t(%uP7^+9Hccy99~9&Zhh&X5R971mlB+G2K92tidP%cu&lO9!L#@;eU}aQ!)z z#?WFuxCfB)-EeMw1TpAv7Sx@RDmea@p0+c=_VLyk#&t1y(!l3-9yzQB=KcrMzvetp z(Y~N}bnfYf=BxdZ4yLLf@D%7__@*cb^_#X;!QZuB#=I3r?AY@giqdhosIf;zZEiAH zZHvq{EGXbkF+*|lhx~L28AiV-yJR+2hk}^h{C_p--j9G#dtz@fR0Zn(b}x@oyS$k4 zXM+IP{8wyY4WSyC*z+ZkjhB={@kx2wd7Q!YIX>vhveVGMA02E<{knai&VGJp)(y*4 zdj~55Ihd7q>2`vJJbE784Qm>qV-c6jXGD*J*7^(4&oJVsk9*jxiqbDS_bh0;o>1f^ zJh`kv`D3f9tC&$tUOp#Jzzs?xCk{#*dXDpZ0h(+srINev`fiN{Fq*&VPo=WN{qIHp z_kI_}?W>&~inPdJ@<#M2K6<@;f1KZ|XM{E>QB1G>49YYB`@3g2_vGzMz-~MaGL({D z=Y*m%k$<9~4(zI=&2=CK2jF7?yJ6@58xo`TTpH%trIW)#FEXI4`RQR+L`zW_WQ6bi zJ1c2b6niT{xPtW;*|(-2f7co|5F(bl`vYC7D9%)ZD0PNhJ`YsV(mBtS>0;Q?@;=45 zj~-w_USaX@bo@bz6(3q7obdl$0?1LeG+wy8Xo_WP+xMCipjMITt*oIPvMJy;4R!ih z{Qi4rioO6c5K5q_4qT32tG;bxkKubcTxsoePyigIy(=xQqlDfH&RqbGrC&!cBf&Ak zIwxv}CYedI{OkIk3y)A66ewSwLzDUp7i$g93oR-SfS3c4fB1dOAtB5Vh&iZjTZ`QE z6O~gh2_*d*_ww;WVu^^C7&ln#^1lUQDpZu4>a50?5Qe_Cvmxz_SBQC$`T5e}}+6&k2kqQ-GN9!T`E4Q3`q(p{j z;#>W4uV28psG|A1&vbyEmhygI z&Gh#s`0y}abGkoT+ll{usE6wh?*L=}L2`70cKR=mcD&!Z^(T5P@*m#QXSC<0_jnCU z+Kjo}ReTy;I$K5-G|Hz4v8lN<%R!~h>Y1uZl(RZj<4AO+w;kkC3{hN@@`-_sIZG$#1 zvs~;)-zY`BnM3X)J?E6*Vrw!|+ZW&(AA_tqa)l!CbxxWGU&Q_+k%R=s5y3botSz#F zmG)4iQ;7@IJr!jj{3KL%S&7Gm=Xg3cX*{H5E3Q!$$6lwiI4NrTn$fG7h5sML5_qr# z-`OIgl5BM4z8R3s==AIQXCl*Q8b0ajM$r5yeCsz{8ONTN*1Tm>th=(=H2QcQh-mS{ zxjW1J)kj4{&d`%gA~S^vUZ z2nwYlIaTYKznc%%@r-X<6w~70N)Lp0YTMYLb~9(hH9pn#^zCo(!OcoMq$>UM?*qh@ z3(rr~1&`eQt_HjmgaRG*O~#102mT{1TSG>}`fUFn3lqH&>6TTMAgz?uu~QcA`kE?r z?OFQfn3^lpplqf69~JIvjs!sLu7-)j=hvcT^#w#EZ~w5jX(@EIF#2Lr zzkkXDi^2UH@-9KcJJ6g3We{)C!P978oi}%3*b`+`Xddr$fb;R#02@m)D~CvXboJZx znOKG*4inGw+pckbViIpv{e566qGd~v=AE* zykX=rk{zq_Mz1XKt8AUm^G(CMqNj6HS(||lI{}7#yYKgeboKksl&pGyHyRzh>aM~G zT9e*j@mtGE?R&~J)vGQpMpq?Mefrdw6YBYpw(QVtRHz-CKafX!I3MQ7+-tpUtfR7Y zExUmV5rI3`d=2;L#&n5YJ+x~y&JZxmMAeF}ME5hl!GO>KaAVn%iY+e}rCXL5Ba7@`@v3Ibr(^-jgox!xUO!nW8rvGKzX9Ubtz%8C&q9t zJRZKa`m^CK6keE>yWcb8Gz_D9CkPd^G-M2BBa4I;R7=?QjLf;wj#l1m8!Lfq1c>gY z^$Jw=y7|Cor#+Mk<_GGku*Oe%{ZEoek(s;7TQvI7;IgA3C6PmgP_#5db^A}tG3;5v zDZ&M0mrJLNJ|E3~ijp1~^Y&rSrMg`?rWp6xy6*a6~<%5(GP z*LACepVICgSI}nlWhi6pxm6N_^L}hlx_(H~b39*JV+*Gia&*d;nv*esQC9_~2;&aa zY^s#G-O`P{g-(q*0J?UyHSGgXD-+^9Sohd-!B_Q7m-QH84ASSbqYPt+-5z0lbRv)4 zESV$KBV;8cu)9#UuSA7)KnY&BncSW^V|o$X{zSjwQQ8bgC<%BmQKXkSyk}&jd9InO z`@n{6(?+;6E0zFB^=S@ zp@)_t^7luCj;pDb&wWZbTKkhy_t~5<#~C#)ZG9rf78ViDs{1wUE4ik zZ6mw84qq#kSYB9g=7XZ19gt1HWAivfD3&UnnHwOhDhUPWCO9iuqL+>`?_YfoV1X45 zG_BCg;$YY0C9k~A>3+^n1#BbhO87wk*V}^(t{-f78GlsNe2+-}-TyIXjV~qsen%Dg z&EtK~04Nc@^?B^f)lvyt(c|>efvm#SHHOEQmtO~{Ppfsyr)1$}MAXjz5d1OFK9+gI z-y1g)0|BDz9n1r*QKGvN(eB;c2Mj6~`*D3_Pm1hED&0zB$RKDH% zIj^b9+4618o@=5fTi8fqWQ&$Oghp{+M&#BPdHIqHcU;p^N`B;=u<6rNs6v0H5Hyt& zdi(wJNJL;-I|q+WoT1dGO*i>HpRg*1Gy=D&O&$AaYt&8!Y@)?3_Prcx==)rHs2p;N zxC|Fm+ARIH7I|6UkX6CFhk}=XR(q55yJ3AE*Gv{T`K@5Fm}O{Cn$R>|*Uqpva7;*| zvKqS&FxHLz!-9sx{AvAxq1!~Zy#2&QrFQ*DTH3&rvhLUl>5C7wJ_gi9AAp9@`m>qV zw~u=lc1jy_7`8n_b;MiL&H7?(4GSue@LYx>sSlje!}z+&+p<;YNG~}<4W+nGTrryu z4RZbe7Iz)zKkCwaY8TtQ+O5IZqrP;nD{NuF#`+TfHL!!&wOkTlHjf&!27hO$f|_t4JrJnxDNh=-)n( zc~dLgvw$~H;r?&;oA8q^^7YYLAE`_DYL5msUHkXHBEtZg{tCdvk?b6FT=MSRAr#m45=!dWVM|34_P&c9^=|#p4IPA53mhOVi##xyGq&1E9gD4-Gm)|!48(uk zf>Wl@E^n!Xevg)<3+C>ibH~X|%pbHo*uL7na4`J>F)i|=_P8UuV(Wwf4NA?PGw%vEIaXvQuIA zPsQ*;)3Ij!CZ(|@BMi+8xFLaQ%xc2hvWqRXH{gjxj+ z-pr7K&*^uEDY_%bt?MXu_sV2xFhD7}&WYBtWFXRJz|??}x?tyP{JF>`n}%+g-qjR} zCE&E9${HrAM61Dx_!1789j0pNiag0e55!eli6L!leV8CvnS@$Zl_6|u4_DU}!=4@L1%2VnH*jE%5wVRZP2 zu{TrG)x^dfyKIUOUjzrJ0r)s-scq5mA?+-oUk$v#{y+ zv#4g;0X=zxVAk`fqt_SU1-$gZa)|MhY zG=%PpP~7p1u3M{Sz}$S(`okYaywB4~oEQS41TCxvn*zNLDRts|kDBvJbz5>`<Of1n1i9W}XG}KAd8Y(HHJE$dUwjPMW-(k;-JWDl^;U3BJTm?#U?37ynh=2ldVa<LOutK>kte2ijkgAUj1O zJ;kUA8=5rOuo2X3`#>#2<#Gmkd$3 z_@WX6iUg=h+H5Mw_(Zs($k+U3?e z^2!T;^FyZra|ogrQRB?n!HZ-uRW2w9i0Y62yaUA!P0!c$D;`$t<+kYE%Bl zdK$l4(LBU0TPo#HS{$-`V)6C->0E;ZYXhjti`yKl&T*v;*OWST8b`>qnM#{}=;x8& zNcc!$LMnE+*e*WN^La++8@5@FbC;TxZdxQVSe985IZW1R9afJ2)xF#kS4_I;iKbP0 zeVC_?xUtp!Yw2-_2V6G#RcdL)vUUytqjJCGIuY=0D1qIvGf~m8~8!ycf32fZn zHOT*q{ObkqE&}kq>mo=5x2Hx?bZGfl*h4i=Jx3kfaA>Q}0&kJ9nSHH4V#^`1?&tH} z2(0fMh4}fx8l7F~{Qc1%TPrIS^y>MDh+R1U1^@K?v>E|!s?1bH06Vmzc!{;NT&tvL zRW&xSmSp*&UcJ|JSQ8#YiYVRATdv<#)9}VCmh^_wdd_hS+92eWLWz4Lt!F;Nt!NW9 zt@=s#Ieq(13MRMm0=%RIP}#I50^0K&dn;#~s_{&#g3U5_PwGh#HS+~R7S5ZW2AwN2 zx~w&(*)RbS@R;{T@QBzHs|nsknION0#x%XlA^j@m#zUk!Qo>OTU~lQKl*S{`Yw))N z-c-0IakLd5o&Q2xR&xF@_uy2fx)43EHq8RT)x(;}^z&%JYfYUfnU#>_>?9Y4!Zf)z zHuA<+v(a@5=X`2onG#kobSOO##s7vV^?>?ArxNPZd`4Ig_G}H zlgzpRj(Wz~1Zx+b5Gr&3YZJljPv z`9sz$keMk-vp}5THKTZ)=yWt3(hd{EiiG6Zdw)5BD6MeCU=jY zCpPbk33EagMi;?2xc)_aHD+3*$D8MHC4p$MTiLA;)`cwa_+FJ|mST*4@xrjqG@eUG z=UPMdo>h{^+Qde`nMJpkAy!1FwHD4xCF!L_HavP;Ya=Z&;FUbr)mpQK2z2}@D-A)O zdxO`7YyV69%u~so(eJr^bVvX?pZ&$$#ftb_25vqbF69GG5t)NH=Mp0^n@O=%t04B^ z&BSu>H#Oy>ov_uyrPww~6`Q zU8J=Ah42~>fy7mI(9&5wmL*k1L&oZati*bF>XWJj?l`*f3X9juauDQaT}m;k64B4I zOZ+ku7Xr;idPoPQhU?JM<=%k|+nwLGH7@+9aCtSVxi_Y(B3jK8BjcrQ{z=Hp5wwsd zApNY{eVt4dB-WohF@7yRA?H=Q&9JcD5I}ClA+5q3^sRGgX&HTmWkS|qJ&k_0H9Q_^ zfnLT7&qr-{iM4x*)O7##-bD+OZsp?g!F#hoUEb@DLqg!@g2i_Wev{xJNi*0 zx0%FfQ(ClxuSk!rcv=5!>xR+!Rl&~_Xap<}>rAI$F~Y4s0&KH(21-}DJ=E1Nv~RJD zuUt|ehSOPY`Rzt<@3Ylq?&U&*hRZ+4`IB*odloI9JM zJN`WO{r-?n___@h;+E#6y;UmtgXCKRozS@E=PFh~Yp-Hz)*!G>rW4ZrVkC~5;T9w# z#H;3_m3JuoXu@eSOU+C7m&eDAym=%H6|%gfdD~VcRyP{e?|92DjtiW|E-1a7#LfS;Nqtb>~QSgi;65)0Tz^0A2?7K*7kGOlw0 zJI!uaAVRT{+W+UxXt%1KHXhSVSX zDyrXNLNf(o1FJG3jP8$Y8*yF{JYLH#S(;Zmf86meQesa8DG-V3LN5h+L%;qTPEC#$ zx3$cSw5HH$3>*t4gU}zSX5mDX%;~c3Aro}{8}-SKKT1XD3Pqp$lg8M;=M6#a*+<%m zic_g4j|CHV|3$>`Y^l>9#Z!Ri{fXUKW2`6g`49V4;CuA{Rn3*hL$&_@Tq;W0LMTck zyAU$=tl6@bFtQZJniyo6l=KGE>)Q zebPP>qD(~k>e$Fb|0l+sQb{mjp+O~X3|qPLWQQ}8=lC|NO}h8{6xD*;T$g~+PKSAj z-ua8&AAf(avMAG#@*85d<(oBrI=G+qf)$}Xj?-vkT!Ck2J}vJeCx503b?gb?E%s%z z-Js)GKjPv}WWfE!Jm;IAn=^?y;TaOWve(+o8AWz6D{WAbNAu8exyuB3%oZ`o00>El zviHEzOIZh3&!Y;EB@x!zVe%Z!#`CYasjI`cgg|R#-ExOe`1ZtbSr7hT81r z^@#dbWmc0#O#c>8>EjsXD7@;^;Z;%^sBw4hxceO2cfJwY#Ku*=5p~2HGQwqv7>UJS zB;}`qpfE}-O~3i4VGQbVc5$<}RWfwcSjvrk{V4W$r`K%?L6yBc1%TjbSZrTbw@@e- z=Yg3IjEk)5T{W&$_q+&*x~@4doW3n__E+r$q)|;l1BXbMb(k+S`aUa=js>KAQ+{Tv zL-50yfnOC6luczJr&y3p__E6&pS zFIC~qgCyoXXd#v@47QG@u5fhUKnG^AMY?=(GsoWNt!3QaQ*8^O(H(8#MlD^f4JKLx z=0TpbvHprUmi$YaPJdr~s`DzEgzf&>#+O%yN$9Nm-=FH_GnB2#(}moj@fwUU=(#`J zQ3w21*?h&ev1A;5Y59&mn!VqiZ4``|c1-|7;V|K~!QD&+Tv{I;2FwdUmTr8y#i#_njVq(pQPu^6Q)$e0zRa8^Uiv-iO5AJ=7o~qx>Y8^N zGG3j}$>4>W$?Un#8*&}kp(gMBm_6_hq=V_3M;A08e!{qtHlf3pplL_b`Zs@urNWa$ zl#FuUdc1R8{a(aGn_`HYoqL>ISUev+BP2NF4HJ$9f*ukcZ?{te4cO5xjyBnSqrzLe z{u>l6VYelXcLZi;aeWznx`hjqMH%r0<(IUg4kPgn;=Fx zV*wfKLQc5;%Z`e-%eX%b7q}{j=E9_ z)g#t06pq=HN0F0|y^sv&j!Z4Z>v)-WeY+yp~^NS2w+@w^wk}#yx>u82SOdj{-uX zeF@r{_!HeE<}W-Jd1FjZDy}~HP14$CH4SbGR?D$*cYD0KBFABBWn{A}hSJ!Vy*csB zo|Bjd`TU9hecnyj+O?3rPA%K4KD>^LCL6DG%0@Va&>W?c7vY zVAUzE&A}W7Y5Vo{hfG>8h9ww?gcCel>lq=*CyCI|iutj)wB`zzEjgvxbLJyd>0eVf zl@-_8flyxu;)U$EpbDW0`s#d5tMS+Q(Fq-9W&GvSZi=o-GueW@!` zi|iizC{T&pG;p@cFKE@7G+a-#mKylm_mi|Spv}gwhGj9JJ@9iv1G?AU#y{D_UUIeU z#m&!LQb_0tvc{WA7+qF8mdEtDHiS4VZaombl7AGW>jAK13S54>EiR-s%EZ;&XgJG1}v0R zv9fwDe2Hm$#V8pQ>|bGCPY*hht}Lx1C8OBT8aX@{p0_saNJw6cnvyLEJ<~0kbEDx) zvFaUtx2>xzqbv(Po2wTzOfx4_itpYOOV^uE%`H_a+irq6C zB=Q=Z3}JeYPj`emrW%At8GV*q_~>L`Bl;+6Cin^=K&9wyjO|iz9LFo41bnFXSu2>s z+0%s5Ot)+?Ab~PBdznuBmJr{hsXx#8=g!Ipm46aHeVt}Z%iP4E)NzKPZ}THz^fDx( zpPxhY0gQEkxb|iw#|sro)j}R<1bU4t-E=Nb`~j8iovARGf+7v{@h0MVE8Q>{8!zb) z#hVXWp!peoLicqbGj-BEah%F{wm9(*UJu#wkJtGl4d=_H4m*_)szK&6Sc*BgH1)BI zs<@BkvDR$4a77<)<>%8R?dik1##XT_G0@zyq=-y7U`{lGkblbE)$f?|!|RWge6ddN zY-hcBku%SWvtd)N>?<(^R}AI#pt@wWj#7h4^G_cxH~8u&Y&w+PGXv+7B@qf#pd@Y@ zd|i?%P8J!L&T2>GHFTd&3-Tei7w+tjeMVU(t`%KpQFc0E-9`gPOczwKb=vRT#wr-M zYoiWfDdKMTDu;#AZui24PR}#zKR{{C806booD`i%&3HYOki1ZJqL+c#nbIu@?5i9u z0RL;?>q@kx7P1g?r$DSEVsgP{lhLOV_mqQF<`t%YClY_bYyI;!_KXtHGZ?TeS{l5A z;7c3g{NnN)HvJ}9A6p3LoQuv@#W9wRhA^Kai{!`DJlV@q?4w5R#Pq!u?$b?C_zAZA z1-^vzZ>6ulOTo=`sCB)!J^#sjVl!op%0*X=2)T*fu`tWre49Tpjg45F_}kpY0qX6j z6t{BUIHd)+re#<7)XaiK-8j0P(DspihrN3i}}M1aXt z(+R(kD3~a^5o~ArrJ1RGz?bp#UeCT?zvM4dX+aR=DLYq5k%sbvJcmPR6v_*Ya&<-K~ z_{^NQoqZs5hUP;a_xEGBHe%o3wXtjBp%LP3g#+q=4Xrfw-uyt4$s>tpATO04+d_C- z;#`1=^iONA@>UhMO2c?w2It+keI+w5KA+*inoTN}PH+^{Pni@!(XCR}aCE^;0Hrk% z{t~M;Cp<7K{`Xs0pj+-6ep}kpv~nW`yN=xYq7yLW!K|W_0geloQgN=Yc zFH){NurcnQhZ2#h<&5$=q zs-J$7US~x9lI!~_ZK0PPS&s&loJ&+nfO-bD>!vaspCSGg0ghi)LgF~@FveKrxpZ3c2bpD=QoRUTFlO~gf7KEpmogw*cF^kn zFdaS9wt4&Y(L@h&sY$&PPbkeZ?oR^N@4Q13kn7PEhbfQBMuJhZq_*(?+0T(IeFBCCR1l{I(K z6kw6YO%5A%BcJwb^uwF=F+IO;RMV}taJs5eM&<*gB-7p+MaLw^WYl^7^^I0ydS~NLjc6_^D#aRAr9vH|a7kICi{? zl#8=cDW<>hSQF&27q|PCiUBS=@Xo8~{uRy=0s(2!`?RTZ7$rE)wBoS?Cu4eRzYJC9)y_>Y6ehIuHu)F|$EJO;Lf4kDDHr~;< z2^f|S0BF|X51dJeL@d|G^a=kOWrx%Bk6V+$TgMQ6ZHDZ>1;~%tS33!uNn=7ASn+z# z5MGwgeucx2iS-bdmm`2P4QAE9;vP)<;8?RySAlIa+P$*e@(AOU?6UbO{?W;#pJMaZ zS6f>uADIXF3dYs#>?)}e*Z-mi_7&2{P#)O=2sAOl9rvDl7Y4R(#~P*k37@Bi*?~C9eC{8RH_|)+3Z5_8$!<-?wXt`;RrRhUGu5 zRejp{;~Jjs3vjT5oU)qX;(n@qxM{y~zzcft4+7^-F0%Lt%N$6sJbI|omhm)jwnIjd z)Xw|h?uqJ_XZw)-#uD8L;<Y?dcm(D1GV01JacBp1>{U+D8B? z8fv;*>!P^wwe_Ldz3LW-wwT^WuU;g6GU#d@S;nd|P(;C2n1ef5R|kuQSU6{{`LHRF zIihL_Kg`n(80<`93skO$s$N%^y6+$9TBdk~>j!_rc9oUDX{lm2om_O!UV=UCG-+^< zxqPG;xGWw#{s~M0B=#sRm8C(k3>p3$paxRj|LgLNu-9v+sJX>ET^+1UP3v<mqaaFrIIvxg*cEV<{br^UIiNnYewaa?_00vB~7izB5K5L*`nIr%9eWJ z=pq^Z@}6sHd5CU3|BF1*&F~oa?M3+)5E4svI$k^M6cEngmRe;K@THrY5Ir`vGE4)< zLrgbMv;2c=Os?eaGk(%tYEn6uyhD?hr4h;+KmxW=KqJ&%it;~nDDTBc$=Rg?qm{%LZ&eU%tSIGHbeB{ z@p@-L7K5OzF0q*57EbMEK`~M6O&y2_L6d_^FfP@|JuuU7CpE~MGslft8dtn(t2{9H znYFWsm5waYw!$(N7z{2>)PrGiRk1#|#FvZ|Bw+V~Ea;{IaYCe)8J8SxRzJUc+n508 z>uKGXV$EPwlwE4y{gRzuwCVS<(L5O%Rzh9!e#r!8!P-TWFfyh+shDgT4=NwPvDbm` zxOgr)*HD93g8&rz-^&y=!|9r5{(Eh&{QZ-ay~#BNe@!R*t>dAK_UIM~FiOA&fzLT- zg8w6bll&{S9DeVS8~b;80X|@`)Bi`Nr;<_p>~e`{zjyc#wf?_~o|Jh6fI^?}smeXg zi)6r(&Gfe&P~1WDTRaE)1OthC!-7Xb07Kq-@BojPGz*fMv?Esp|0rHbPE9sf=HAo) E18xG@ZU6uP literal 0 HcmV?d00001 diff --git a/examples/img/graphical_pipeline.png b/examples/img/graphical_pipeline.png new file mode 100644 index 0000000000000000000000000000000000000000..b3f081590ae31a6705d10466b1cbbf4818441324 GIT binary patch literal 21525 zcmce-by$>L*EWuXh?I0lH`3jMv~(#UEiE~844{BCLwAFK3IYlef`l-1gCZ?MOEYxM z??P_x=eeKvd4Jz=e1Ci#vS(lW+OgK!Yp-?gb1+g%LkagT7)jIHiZY!fMgM#uQ4*S9a19(@frS@0>Xs6~^kO5}{ z6&e90kYfr^x%s)z{g8-(TiPjYVe1%Z@eX`N%*ZEc7k~3s%rfGpf=9s+Ts<#r`f}^w z?53r#bM>ZzLqdI1Y3=_GRIu)m@;&i9J--C%9n-r&c^exaC@Tkl0VO650gtvTPB58$` zuV0{`Q1QwF&q#3iE>G7h?VHSlvE|K7cuI#6O0-M^{7tNFsK>ju<6f7`lgaVeeR(r^IksOT6h za>m@6&eQ8hz#~ZJ+U7=C-_S5H89`5OIx|V@hsGq`)K_;;*%dA5#kH=xS=ih(0_FJf z0Xc^RP{}Q=l&g8U2GkNfbTo7Uh1-!x<2!wT;z%pt_ud%Nk^K5$IvnXihg^(9LVyQ0 zEj>H*4B3bQ-k;sV%nJB;c7X4a9eQWIW83(Jo({jz>ucPeL#0<)WzK)NdK3pv}i|k;d(#gc5hohPA zf@)I*=tO?t5!Z zjW^gsS*H!s!5ZhTm#$LzcaNiPxxa_je2(oR<1Xl@x}Qb{>OO?M7T?REXm8`St4ylI z+%OhUJIHm~S6bQHNOzKdN6JmrT2wr3(i zml8hd;w$i!LH#wGe->%Wv9lO_`aExD;feW0W|#(D&J;wOK+|KQH~W;lmu#zT+x&;{ z7DAfS{M)ZUYRY*h+|MBcZ3HPNEnd-Nv!L{f!E`d(k1TdJsE7{vU&0~nZD-uHCscS( z0~iz1?Q+8s*IMaUN+l!;l5wKn85y0&aM z=bDv?JBs_MdK0kH_z`1g7xP@E6PBd%6?&$m;%~(t-dB*}TC{!o{DCN?_L~#MOGe|e z`|Y)#`5v1jqW*GgyKya6tmdBEh<^OQG~AW9Tn)J}No6{T`~68MW@@(t-p44w=u!B4 z2*WG+wZVN_4~@lJ_|oCwt7mQWtUtEs)^L30>F%7>5bQ*Xx{*XrGlRpkLfYt^%Bv6$ zIGq_AG?^|41!C_g@ukfOK70DwdDLJMLI3%{lg>X_snJF$-jx%298E|$F!Gk^Lp`Tk zhlQL+Ru1TMy>+4Q^b&qw53HU0g%Fm3k3q%L*_iY!1IbNP)u_HT$+G@?$30m6WVaey zKf>+O{8IYwiHy`$aU`L_^)|#Em#y0cVz_9$taTpFyu)1K9E}d&bmY}Uh45kI78!eT zn4YD0Yq5osl$DLQ)@U^Ra-NX`{H@FxrrL6eb$9vaVe5g>ED@w4%I({GpY@%9SF&P0 zQoIw}i|z`*XIVa8R?}qqq|VQ1c5{{mnjLX?w4+rsv9G_QAwo0!n}rK=5-0fNLDGlQ zhSQl->ssQYj!QTh2T9kjUZO75)mPmZWJ#PH(5bv^X)qI-JM#xN4Q}!R7{*I6X9}FR z#Jg>iGTyVn^{jgrEBh&4!+3TSz{W>!mTHW}2>d9B?R-%v0lQ;!M6cah7MxM~5 zv-Zrx%hY%^6CW9uJBK`vQ?s`KztU$b+7?mDJ}(jF3)1UnZZ%e>gI-Oq_C?a{NY36l zcD1dL02?k1#Z?nTCK0ByMh>LkCriU)Ifn-`8oFZvbKZqfSLn=;UT%0uOzT(zvp@0U zjx_-%0lxXb*|`cg3wK5N1y9L~48ot$j=Vz`KfM(mVylQz_1*u}(7)O+L)UBBoB%HO zIkzjQeY5&2%CDlLevis2ghlwps;ra%i;MRb1_z6GgC~PDX@Vj%xGqi3L`JHspT%9+;K}wcG>GWtVo$9^HE5i5 zR-RnZ%XWtTe7|~vf7Nhpq;Df~X$NVcj-4V-1J)WC>a~ zO8JdHl|RAo%dc2^&sBP=CR$6Hi2#i)2>8J!Y@#c;#8#kBL!7gWg?&A)Pu&OuGVI%s0lfgw6v zZ_gE1bjFP?+||8ijSfv$?S6bql>q8*{i0pGe&om%q|QtT**MvI{YbeuxLOu1wa_>E z_2sJ>bb0}k?OUQvC7%P0m#vl^U0(R2ruJJBO0U~gIozmEyD_RsZ*f@>GzWhr+$wqr z;dLPx4)zSBZm8H_y~|G#+w@IPx^P#l_M5LWQ*&-EBO*_PR{em=m~~X*e*CS zqEs3`t>f)m+^A|Zl)|sX{h=K##LBFtn@7QVE_Fg{7?>L&sm2Mgp~q(Pb2N^}w{z4p z+1D^3#hfiowH`wWr<$FT6V7#*?J}jaQ^iC1-{?2ax6G5wJX)XWgjCPv7J;KT=*6=8 zVFM@R`vI{Kbwok8kY-b|Lv~R&$GrQ8(}%exJ>RP_*W-KxxGypJl%H>ljb3rEQShBp zuI4tdx#3NJ9aA1ESRJK~n1eoQ3O9$YL`uVEum$uH_Cv2AulD86nmU(rR3#+HXo4vB zjaX>3g_;h@5quoSGe3UW5UezI`u4jghp#by)=AZ~zep>=fxQfmRI-Kg{v|dcK$|fE3nR;^?L0{j5^c7jy<94F=uD(fyMU>xtCrXUgf7>22b9R9ZZGHeU~AU>Z0dM`Q@*H-ktQFb$8mm;qj!qdwy=#OFw7ppC@(%!?N7z zpYGVv)Ak#o2*%^DaG^fc`h_WJfL@cIjGC=={UJMoE@m4R8W;V7bib_$IJ2JC%h2|) z?_c}ZMf<<-{;G$JgUe*8^>xR=LJGjgj41aVXumD@Ui%1zr!mNxs%!O(OrAO|PNFMe z4j7rknC**2V~^r%&bzxD*@eeW=O{2Ko%6L}|B0$RDA z1+RZaOY)iS1FWJ+SU}Bu=I+NE=GKf+(Y23y^|sHWoVY|%#-wDs$+E@$^kV-5x9z%G5j{1@?-L%4gQph z#P+hdw1>G^*T8{sF}40i%5oyq2-_ty&VEUpn~V%vR=t|)IU{izNAtdZ(I&R>Zgu&h z4bId%3FprGGT!>`D!`W}+@E+G2k57|$`TBJjf}^)P%GGnmL1;iO`QB6ifsV zQp#WL6b(!MlCt9w5m|bEwbpvOlCv_rPWv4*?ve4AVp{xp*|)yq)9yneVgmtjvRwAN zz58j0sTKAWB@p6fs1D*y-Ricm$kH56(8IT>fUsE_G)@ML^( zX}8X~=dC1H>c^GjSE?C*ZHiXcDvMV11+6W>%P!;Sw{?q;&gP@sB&X6Q9=!B*=S8+Oa_eFt>}31Vd%b5; zcrSK{phalR?SdN5TA%?Nc5Q7VQq+{9XqE~7pd_7?9R2^kRrjI zGfQ@BbiX0pmFhA{oods6*d}C>H1o)A-j~f0M0MPola-d2E#<#6cet@Sl&2=0D2x*q zr!+G-*bI)tYL0&p1J0KGCRy?onq_I?4t9IelO@X7)pLbeOe@O0P}lI?S~Xo-31K$e zg41Qv__^f+5y-~-)wzC;<-zIrNx3)_TUfjM8E6c?;=33s0)C2Vw4az}K3p1m81g+| zQx^s<+DdEiOYIhJv1AZq;AT#*JbmZNH52{D6{{d}n|^92?5c(i+W2CLTH1%k6fUh&{hx#N&lB%_Lvizr>VW90i(oo^*G-#vo7eT zQ;EHMB8r@}5l=dL`{Q`0zeHm2D@BsDXS+TfYW*{lp`(`Cs|AOHRYEc^uZoPj0o6l6 zq!IPf>7|w2n9^F2-Bcg#nXaNN19N>0(o*Q}hpi|GXRO3gP`y|B*p@WJLQWI0KxEgE z(jQE49DkQy`a=Y*e#a+Nf@32@4HHL`xKhWYO>?;NK-c|A0ssSwu-l1;4B)+bocDn0 zl}zba2~|@wwTJz5-F!)}xqGxtNoBAT;^1L#mr~Jcih6W?P}oL_4jU!-aRc$b?p*vX z>D0RpAER9NAl;3K#`LU3uMj+!%SsYKGmlw9??eea54<3+%OMdm)fAf>)n zJ1P9>(r}%0D*by;m#UYJrmOPFElys{i#<_;|2?)YxVT00NJ23&Sq1*X+NfkOQ-0?Y@BLM@#No(ffYs6`#T_y+i zZ~rZ#Kkwgn<^fq{QG*?LbmkF_io_GYk3&TGCa&9g7aLj8NxYL!$NTp=qk{T(f1ja5 zxufkwXC85@0r0D5d})F~hEbU<=pRm5-|h?$=)#zPchZqN!gf2yXCuoWr*D*G&I(c- z%l7|$>SAyw-7S3ZU@&&yZf@1dl^|vMW5(X7`!_G~C!?3V0yCZx^wNA()oDJ(7aR}^ zqtZe(@Im*z20Uo-MWq6$Q-`f)k<@@zaNyM~Od+^dZsZ zqihG6M}!5<(jkSTZrt%Q1PG! zBSZ>ip~Dnm!61(hSFny*prW(%%a8E%iMI%8SwO(0!>sGY2J80FDU0=*s( z?5OY{CAEu9cV`p*?kgK=t(qxn11orbX|Clwxc!c0pWZjeY`q{Eg_iG%#=4jz1|5JcHE(;xHqn|TF54s~^R+$FI?Ki}PH`Z++_yPJm-+Iq%_#EMWl&G6IZ)zhD0gHtH)tntawx~Kad)_ocP};ra@M{Ro!89g@S*f(5>kHZ2!v0*= z&G%D$z=;9OGY$}WU zy+be%O1AcH{Vqx}R^rLxyld&LPj|Mw#Iu&rx6j)P)_;J4)W3BLwB?U258*gotIn`% zAuyNs=yYbO9#3Kt0IO?pZ=FVH7oQjo>Bt{8REdeg@pEBnv~=WRvLPmhr*2{mUbT}B z?1)+O_6wIc$1R3N=WJaUwdj*?cv{Q~a_OPDb_^Vun= zUd%u40JAXO1D9mKp{Tqchi^CV0E_#KPPnmi%*ZC&wPR2Ita`w5HJ&4Qz_gFzW7${e z>F9Y-ceE(+aKU=S;(+>t^1j;8e#!lq1BkF*IHyq9BiC0?5GL(AGk&h)bNm!M zKejXO@wX{_a-u;)o0?jXB=eZsI$X$~D7f%>ShnO9&B_9VU-s}lnn*<|cbAXHWAf+V zwc)_wZB0dE5z*Ji-ODsvqFuQKLAq7qR&V*_zmjVzPqjO|i6PjZU{eZ>(@OunrpWHn znWQYdQgmgt_IPpkitah~dzSn!14}aDgQS;Y@Pu?&O$eS$>4Z8twPhEi8MtoF)x}V6 zfPcnP$}z=!XruS^l&lE-@ODx_y*=ub+3mua{Ax7=v?^IgywS7Ba+E#jvQ=0+{5*K% z!84Cxi`j^^n!1o}`Yqp&LX()(ULo!IFlnJ~uPO4Q{uYWbq=9qCEXa3%X&tdT6y%*# zuawQqP1MrtzKUm-oZ8noPR?L~Lj+!+K{lSHUpa3=D4P}Y7Otl013qaYxnIQ{SAv>q zX}5Q>DAxK)lPOQV7dppw3+!!CIdaUny8@S|RzoaZ(@C!4DZnh7Nx&396tVDg&1BWN(gSE?VSfFThn^*3M~w0Z>D!-p04(-1#o;{_|0X1i#IKMH^hl9BH&LH@$zFjuutq zY>Rx}J$BSyCOED8Kniae&JtnZp}^(hwLF5C<{Ru|JR3t=B79s1JE;y?Ds6-F^QJNWz%niz5Vgo=1e zpifJ>cS_y_(?ietQX6`zz}^P|QFW{?yqOq)KKG!Xs46DVXC>Y9jS2~#PNoE7k@86= zEUIa4!9%W{>&&cH-rfWdYsjMkVbo7L8wWlq+UqnNOHcl0FOpVs(le9X=hQvBkGln>!7+`)W}rJ+M>XC!_hI zB>cQv1`?6$7BZ;C7|sV~DyHg2i$K>mI202KD+w(hUGD+4 zi;re!?`Bu@Pe!&K`A%!t4%-&BqGm09jz!}#)R7%%zakt?h>tX$0P*_?*v2Yx&7 z)5T#vAq0#6Su$ly5{(S8`ev`h#xfTVA=sF@muf}-e=IH}22CIraOD>e3dKW>y?SPo z;WXhWIXXY^VpUDaSlV;b?Hc;OyN3Q~5ut{m_t`jmdcz#W;ZK`zdmeQhuIIhoi>B;y z&#q#&&h+GjpX@~ZyV+71ArW8I02_@FObBO1Hk~Ubc@OXrUR4O(WtxkV;H5=7ODq}Q zp1CzJPf^0)JpSO{_rgN#0U0>NbIy2aRs^*6h#hY|rgVRh2Kpou|11n&=g7Axk>f3u zEuQoK&a0FR>EJY5=swduWo$kw2idB>_S1rN0d(n46}C{ElM@@R|JD^ZEAt@yxdj_y z;A_+nPhRR!KYdn7tGW8KqXn8#59kt~YM@?hmYu4cijw}pDDCk(dLwiBzL&UiQrL%t3!B^uFMbQ;K zTJXn7ic)eT0^4WYB5*vT8a{`?mNCIO`eNMWxt~?j7M9S@GT|$WE~hbYUb_)0K~=Ls zw|kV}Z&r~Ji%%1%j*o4cCV#d?yUS-=L#f-pa=D60z0vkD&%#}-Wa;J33(rh>8w!7EKlFC!$5G;7w9c;(n%o-%SrZg!D`gmd~{663Jl#x;zp zqg0~mnl6thitS#*x$x=8cnj{>v;JXI=H}HweGZ`h&O61#y^;0zmkx(4WazhNl7=!B zXL#cJfGzDgtv4+VXgwytkh&5`Sgd7x%?AH0)2O6-+TBwDVXT?va__dLddhHk8DQ9? zg9^lpuMC{`!47%Ay`weM8n}mXz|vi!)er*-Nq9m>@t!F9m%j+WzsQ;*zYt^V0#i!B9jP01NPiDgkhBV7s^sxK zl>>(DMl6~%Na*Nf%l=-A&noE4K~ybpdGza;Eonw-MMbIpP903wn?^z$tqn{j1^;eZ zJXi-N2>wX^I|?VtrN(7&eJbPvE|(+}8ThCf3Zoar?9T(ige&uAe%NyCD8EG-kH3s* zf;!+}YyXNj$#okjZ4RPQfKmLOV^r>_-;HC+Eqrq{+2M+8{J2lrP0T`_^Z;}iVFrVp(g-9{}hbv z!U~q6a1hbD7>Kt!X`%+@<3#&smD3nEvwtYyeV&S8Yx5J`Oi6)J2Mq6k=NUOEF zp%^e$b$EtfjPhA~UjBeSidin^lrf#;1g>zUtAzZV2wD>gWNxc-T|1A; zx;=Xjf>II!uF>BKQ-3c7c*>GKJ}6Did**X=}~3~?knV4LA^ z_p}J0UA}w$Oup`8{$LQ6LNWU;(d+>q3UeF2-q__@3D{7?saI}Ze+KB}xAVwwEzkpP zVIrt0<|CZ)*9CBcS6{%`_iLL7V(a_rt-cOOA7<+MnBk)NSwPQ|F}hOeI6xE3&&>`r zkmNIl-YADRh&&l>`@3cvCF;Hqg&ZW!wICBg$q~(!^^x&ohljM*1sjdGzg8lBh<+0Q zex43v!>kx--G8wZ7;c|Irtdfp0${kIv^<9b$s`kiohyvT>Xl3#)^3X{Mxt65)30qv z_b&O)cG%GjCH^%c2-tP@nUOqKz*M9{-909s<_U$DJ1rB_tDXXR;?jrB zaKB^Fr&j>~rj9U?(a>zZDw+dTW}RNJL~O7L9Bv0Z$G+y9R~q5iZhr9Ji*5A>Mf*vA z;L!hXB?j-EM{UfWkCOrf zv+#g7=KQ@Yyk1Ih^vCj3z*YHRqj;1N$;MC?G`g`@qsm~-H(xa8-qyl=W zX64Pi8{%t`8@$u`neT^efvE7`t8ptgT`{gByoQ~RLuj7&X^jxe{?=FDUJv|j zblsCfTDM5E(Nz}si6$?(%C4vHeF5@=;O_om zdxF5!;H}`(fNTEpNxEk_^i#y}$KSis7{;)j<}ehhoLAS=C0BiFr&l5S`2MJLZqLQS zd&T-EE6>MX<6&BLok&4NFK`3|ZqzAHA7ZteGn4RU>AJ8h(-9{Q7$Q*0v#|_qOht_` z4$Bze>=d{dAa?v)36u6pBwZp%1Zt(oXOm6<-j{L2sCgpxV{`PWC^cBVuiv6-%3c|` zO?5#6&Grm-;s_M7J=o3yCE=n&8Je;6l&um(z62uTcI-weHJ0Wq>F>Kn^p*=CD5nYz z=~4CCWv$OknFqW>m=UL57pDw^9(GhinkmCaE-7CWf+z{V4y&-$AdrMI zr=-Eyu5bblahpo10?$NA=dfe2dGtp#+D^W1ztD-x47CR1BI*PGjIN8#^dT1w=BuvF zYo-g)e$F$B2?R27)|pl5N@ZqTCF9^-`nw1V*u`kQb9VE7#Ge-0rssVRc6CmCY*)vmCOaWp&+A>> zW^~^*QJV$~I0kNS8N=C{j(D3|`99aR&~97b<^WbATqw=$CjftpHB1k29#+ROaR9BR z7hUVX2Jc^{gj{4lgDRlcj7{FB@*1#@+3XWKiNdSyaWt^*Q0`Bm0<%UUIy~Wr{>R&2 z{Ejv!D|v{#?^s&VxGX8o2;D7;EnDcsP;sa_zF*}JmDgSom4g<7+C#?^oNGx1|4u}?_N;}9>{qUy__ zq07`VWlqt^U7H%Std5dX5!Fah&KcC8?%BA;hXl3^tY<`~#Y0luj+;$k>HJ23jM>=U zG)nHwM+Z9N((sQ-9V#Ez_jgw_2w3g8#78YbY8~1~aO}|~2xgFmX$67Hr4BQg6dd{l z5v02s(Wr6l00l0U`>92dqJs|T>dVT&RR-Anq%i}SNk zt5)gD$$mywbOwzFl;AKO2-KSA)_FFqX}oM=}Be?v!oL-*hRO^@yYWJ!_LH^Wea-X8$O1??96Vc zsDKc(eq@=lDB`|}n?M*dLTv8&^ttVF2$1?>lA!fvHAY>;l)rFUz#%~^qYp!*Nfm} z>pN1vg8&gJ z^z2q%T!Js|I2F(ZzCu>aiO>~;;#r+hC7g(A2IszP z=kbU#66_O6g;Kj#`wnJ<){F9p#gB@{0|u?D%Hk`VS_RZ%WjT7o7tCyZz^e_+tgO#Z z3{vIeVrwzzk!XOe+w1q zZUb|<*KaCx;0CKroiA&-^x_%t=M99n>MJJo<(Y`}vG~df9)eR| zSgWH%kraa;q0^C^L`8(cAJB&hp|VvFR+V4RA&3`=@*35QAVJHRxuZ%E1@=T1r`@Td znSr6eK$|jL9!2M6WyB+(`!jMq#p~93l?aQf(d!jBR2Jj!JbbWALV)#qRB!iR>*DC0 z0Snzo^9bY5AOniizm{)Pxn8fFEmn>J&s|yLs`o#Cpk{^aL*+|NLuESW3blJ<1QRC5{>uWkQX=`)5!U z|NdWkgrxjmDbLnp!^&H8mX)ppnwF`01(*HHYX8e>Dx-M8fsM1EHQme&i>H%XQ4$Zq z^f%k~znPD@fH=HgQMp3v*;~B-e@i5Gu;UfKHy8auJ=&b3w&@*`f1kQz=I(dQ@GpAz zYW%|ai-~G5zi13Qxi7m$2a?r5;c~f{8##33iWvg4={4h9S$QQ78fB^m@;0o7ZStNP zr1?e&i`W@-C8Pfz7l!Kea|av*a~GGJwpjPJiK|bSRH_Z~&Ew)vVJ5$0f@vuhIsS0a zCM`-0zVh$%ljz+5&%g8q}3f@zQ9r*rY~-7K{Ih@@r9DTA^>aKJ$rj=Frp>`Tydl z=8<*nUjm5BUf!Hg3+rSi6;?);F@ydKcBnUcm+p*#Wygx1$ds>|UP+1Qlt0na?k@ZP z<}j)G#NJpiSOevpY!A)Nc7>&ACpHn2(b}UQP~$zOS=jFr1v|mIDA&ES-`bgBOH03w z5gZ}6W7f@kTv9BW$^DJ}zd7%_hvXpne@@w@{LX#cxvuWJ1PlI%cd}#c^cRhTvW{ko zpvXS;D$CA^N6r~o*?3))8yQo#2IS029_Z@M&{vdD;ejbu$p4!!sY>2GHeDS0;u>U& z)Fa=P5`J<8|33IB7TdzdVlh)~5}Q_eL5CbPJ9#RI?$7^?5ST-CSpkB|hr?0C)VQcg zIC)pRo->qh;uN#JdD2>7U9$q7rpj_!@^ZZ9ZC26}LB>6y%!j}xz0JNv#gM@XNoh!rz4JmjL#w5vPED_se(T-P z4;PJp3z5vivb|2|*6&2JpN~GB)m%K8q8lXHpL0LF<R^pbD6j@!&O`j_Ls!f%y)HoT&DaH@#+B3XJJd$VSQIBB|5hw4KjT1wD9(bI-m zX{-TeS$3!>*_Otm9ZT!S>500=w2g6%kZRvf3|0mm^t9-82&%otwVsbzVHqaA>fk7T z9>QCXFj6CQsXjhQZYkus16gH(UGB~IyxY3UiCE>mljPz;z^Cys=0w&;HTEeV3+}c;t(Q9wX>^MMkVX*cItN{ZCwivvyB&B-Z@e zml3M3MVm9enh#pvbz{+8!A5m2lNu49N%aLYB(I=kPwGh1m}|(2CyvfrM3*hDKhe!D z?2o?AOUw__;85ywrSc!|U?F)IQU0_*3ZZ{BY$Ozj=*<2?%gQ3glI5!^Nl;p!4>O`Z zAlWCeW}bOPPxVs?`14Ae;B3|6ox-jjtNHK`=27SEW?>Tq+fk!}J|r9bu!V=q)1n>} zVPt9n9!qWt?At%`t}yf*QsB~;ee>~7KrXw>9y;?JjKG&Xz}|vs3rEpliZjC&4w$w| zlGOm45Z57pi6fGf2v!RYy)5A8`_DL#&Wk^HWjd9OE#sZkO#7& znfszRE?d$8j=ipY>95^U>R8^bbvLB?`&D=_P)pk%=Q635yZFJv8Vx7prn+^IWbK1LNz zl}TOQM65*g`rd%Yqjjv9;(DvOM4tDr>n)K-wDS9ygu79roJ{=*cR2~>1$S%TBB*cK z?8LgyJH)YP&H6nr5m=!?J@fK7N13;AC6TTPjktivbck2Z_M(=C(CkKv0tn(73%njN z;vYTdF==6V!iS@`!k+-gG1`wARM44OS<8wfR+4~>6sTqSgJ18O#cb}1b?YUw&|5$s zS3NJBm6(X={`9Gm`Fmb6*+xyGRLDVRgvDO~LZ^XfE`WZ)|M#4!aJ;aU{oHn_vTpd^ zdqM$Ss(^0Y5B5C6{e;N$o%}Q^Z{CEfP6J*T_4(F(Un`IMqK0;Fo4l{C&dikls@j{p z^H&kxL*cAi_d5O{h_^9LQvk%_C+mI8_~tW($r;;v)X~bFK@|brnKj?s_iN5yPr4w& z8fz-=`c%Nms-yUr>A?|@u9)*(irEC+%rL+=+ulV&%|9#3Q-0B8`aaK>GD^^1#qTCx zDKP(eg!13Qp5lG)1^?j+He8t-B=q{um5l7A+L@sSGkq9IEVb#%RH6Z%5M?7#2Na>pyj}jqJw#gy{J)~ zKd@(SYAT#%yuFdOz?9}1-9d+m$>_s=gENMc0E&|xRdwym7k>~Si^k8}Z#A`eaDhn6 zsALxv{Re1W;Yqm7=U1@JG5&#DxU^YX_(1_^hl%&D)ebGGzXNR6H710A;;5hb zwI%R=IMq#GSSg%6`Z8r*{EuAtv)7$}mt!9VUBQt4kM%`+ya|8VHp|Zn{T@vI#a-ae z=kyxCf6vZv0_jxAT@o0{--8rtq|PPAsTr^L`5(v@tagABTK$G2s`PKn&X=2_CGC0N zv-Iy%0HF9A7@dud)#4EqCG`Bf753){{N0^qv6O#zCr=*OgTJx5t?PMm$656o3AvkN{%@EJ4t8#0LPI_Q z_E+N1{(@OrlC-=pw}Y^qj-WB;cy23HuH zjDFGliYoL`K6o)TU)1bpbvbq?zh0&FBXHFkYZZc3CLI}Pttn2JC6|@7iNhf*1%U|BUs0oXH0IJ0Dvi9O;6i5E30lZoo(lH2 zD|=gml)aaTVGeT%T0z0@3I#*#HnzpDXYb)z0`0n+0ed>|lS^Q?ef5y{OW{d>N+01b zv!d%p_08^p7oQ8+Ox?5@Ow6Lo1!JoD_YeIz6!tO~eYie?LQU*RNBl*&0fC~@n&=V& z_48&+1+k4?md`Mvq6$Ts{q8h^_GRozOPBr1cBRHi6!r)u_ZKGoO5;a4cTr)s9!eV8 z5Ej3s_P943Dgq~WMF#5wh?)!8pluj7pT)`qV~(wtq(>4e*+3x(!qsLO6Yb145%GPJ zVk~p_BcsP8#T3TA42;Nt>zNaenMQ}2B0mEP?i^@)MUA3V%m^mXG)G;xa13>K2a$Oy zXg4jD60Vn;0*;w^3(rDQG!{?~SsP`{rYbK(8l(xz#TCPWk;;GzvN1?@5e z_B!cyn!&u{1PqM>T{J#0{H;zCpH7}8ASk}o(uM*IuzcxWAwfWS!YHTn&t;wK^AeQ? z;YqlXdI#LOEBd;n-3b9HAn4xhzZAw*+UxfcDNg;o-xT_&dk}o0`KV3V8P31IvjNR6 zJi!_KGT#EoCm3?SUgE&|dbxzP_<;>TPPr3##VhvNCSBsZ(+v5u9HML4Mfaj`F8zVy z4`<}bO?-r82PC}*N=pdR&zpG1212}(KO0#5ZRa}@Zf8w;3VRtMLyum!gab>2`t$Fb zo!+??ZOJI|fc4D-6*-^13Y^Q|UZ>PNQpdnCuw*T;e;9nd`)lvJuRR2JL72P#?B4HD zN?0BN9(kjJu5BsJGTPU>db<&{Cu2{}x_4XQi@TyW#e>YB+0$|`06DN1 z5?sp{c0-Aze0!n0qZ{`lg6JOpABN_J){$5$#JR*Q)y{kPUzQ@6ryp*S55NJV!rT53 zH#clgnz64&=au1j7Xd)Pltw7txu%S6$Rj;H1>!ZAf?UPy7e0U3qMN+^gNwF%SUTxT z04tEqGznsX?dagWp`*&np|;G!nzM>hk)pTu{?IsoaX*%Ys~1+z^O66>giD>&SO*i@ znsoougxMSX2SbpTe!yi>&o9a@^e2!+>3@@A&#a3f+%E%xXwBGx6@mHBG&SoES)! z3mti3{k?y87BKsak8e;z4x=-eBPpYi*CJDFs1US0G8>&{rQmx;1yGCg|>r z5?;@y#dNH;@Iv-4ViG`LB@o6-NTw@VCJy4-UbQ`MV7v$PZz1t^#AZBFp6nZ zHUzy9FnK%^RHPUDopFfkA0ioMx02}d6b%CI>*g#7GnB!RyP*K1um=|jHuunI0SksR zftwCrk7I{5QiH_X$DsACRIUS%i4-iZ$wlm7ue`F0J|~2L;tCB_MG;6)AWn!ym7e86 zf_i&?-K!qE5l+=^1K2-c*YZn_itWxyE#h|b<(U}P!j3z!w~Hy>Cxv{fZk$fkAr>N# zv?M0tf5G~Ug+dFP?w>lz9xOG_7lD+biXJ*H2&`qXhYSG!?fF%^^Tljx#>0e$9*oT4 z$>p!+4;x&jiS4|r+VjS1n7(?I=8PEpS~7b_SSf^hRovrf^WlAdAMIliY;vW4CvdnE z;&$)CM&1iNBQMXsPZ>&3k_*zOc6kIo z^E0HOX|s3Z81~z2Oy0~N*?5bpH@JEf&<9{di#uKkRdB7+fSNV$X~Oo|#|)WvF<%=+ z#81=SF@2{j0X&^ge*`v++2_?mQHanNd?X6fe=T^o&wFo9l3wpus) zEdoSy+g38^h2PA+RlnMWjfQ+tcO~)Pht-(C#Pnzn^8(a>v;>^`5kE~&4{`X6EpJxp z+nsoi!iX4zwxAI8d%G^W>9<)c5&c^H4j=i2VA{_&T=4LEa5imA_$Y;7DWQP#x%uu@ z#9XEgjP)Qp0p51ihJjb7U5I9yC?tjDHNXS~V5mts`|{fm;_`$hykW0ky5=>uTbHfF zg7|>*%ajqZG;TWcW3Tl@;w+J}01=TvK;H@^}63jFCo? z_9u^5*lU$#o(pVWhq6p@X zGxaNyR?1ZyBc*Z}wz2G7P9+xIpaGEITGRy0@9VC04Im~b0Rc;O=9oIfK*onc^eRHh zCiRHO&H6Bdo!<*6iD{~WJI_!h@NU){4m^|o-&J5GkpETlae9rx|5m}Y%ugI>m2XsSZ3zA@&jhyDBSEf zcQ04eZGEs~HT1gb;kKOTb#9Q3)t)oCAf&L8GQN4_VaIE2HEfVB-ao$vVoT%mA*7 z5jH6C{)Ml#KO6_!*#CDv8RFsJtEH5b0p-OKWET-C%KZhnA9cane+vUUzZAkI+#~aH zVo3{+#p4KULzMm)jG#c}4R@DlT(eklB4>=-%>9_c{hQ(0#M9$O073;VZby6?vOj_I zw$CN%^uqf-)X5b?ReEcJ#n?}+Ydu98kC?=`?QgG@)KoV-a2tt(DoE05H3^90}%m?L&oyDV^+XgVCEgyaQ>IyvGAa|WhMK?h;BiOuEjGWw>fj|kAu zq}lJqs7OV0`Y>YQR{EfA5pX7V5+*88!h%EXD*c@BCZFA23Dacdhcc!A^akkzECi>PY8Vg746!ZFRXbf4H*` z(BrHgJ$vYYpFisi`qR@IR!Q}1^$ z{!HB$EtZVrN*Eu$wb=+N`G$|5_SDYJBP+p>R?UdM_Cx10djzC-Dl2l#wh50TaUsAu zj4>G1dQ|j186d=`g+_uZ?D}Nr-p@ zEK`#3bvJ+9hT_kQ`M-KQT8k9Ge)s*A|B^xQ3+C4=lZ!1yo>P?G@Lkw~{%NZJiTe4B z&0G{{^LIhA=p31!&@GG%u}fGXd~5%X)E0^(h?A)J$JgGr&;|Yu5oJz-m~P>Z{78!O zMuBR)h5PYqL6j--H}E2KP_h$vpwEb2uuc9)_<(x^UjNTG;1a?eH^?Ts?7jQD7h>w4 zH8za#eQDr+LXCt@*mgkS#`=g~#DZ^{5x=l-@S?)D_XWYkz-%NkY#3c$S%qjvS(igS&nT`;vg}0 zCZ@8FEz4kxnR$PsbIx<#|MR^6x6dbL{OHWxP6hgsOvxkk7rFw{z&&r7RweR!haf{tLjwO~i@$C*sK6BRX2B)yp^e;_#^ zOivz*U5q?o?|eG+o4}cxa?vUeV)V+qcjNB%DzVcCz3kbw-T(Q|{a2Soevn@)%iIwZ z5ELBfiIiP@q8M-qVEF1Z4{`?_zn6J9^rmF#4x6yVOEqJMmbGDnbtz&&Q6#G$y-MJ2 z?`FMM0tXuld+SgTDI=IJ@Z2EDRB#FgoSqI?AYAjL=0;l6sUovGA*pAk;r%%Ekhkch`f)A%&lq%Te&5zALIalRqvc6Ii z$8r+UxWt!BpIsQm;*z^`$I!qxp`G^~{l~h8TcR&;3w-?Ynofg8c@EOTZIa;M)1sBn z5xf#)Lk>DRYUMP2^$^a=$J;rUz9m?k-T1<4Aoor6p6mlqL|$miY_u1CUM_JM7y+G>?G^v=QHh9GfXM*DP$>t@$s)p>5ADw`65(wOFu3*18KpTI&xnM2JNS9CX(az$C#`}4+ix!^w9MeY*qTw7vZ zc|E`!o0;p-XFFV@voh103&z_p%XlRe3c?t*twR2^g(lB@Fc7zqqx}46@Mw?Fz~Wcd zmnNZ1EyKV5?GvTkLegMwKzYj-=Bg!8zDkHtWq~-HX?eG-`a{fry0zxxco>=5jH@1ce$TU9-f2#&g69W$F4d-JJLp50ubif&o(;h$CbYA} zJyv;Ntq_s$bBJ!%-ubr``mU3#?6ErEgauVJof6 zsT=AR=CJk$wd+pzd2a{YssFwqcQZY-bEe_3E{ku1lf}aON-m*)RQ)F6W| z=2R}Mf*7M_kz3(ZF|fSjcqjx67+?J;BSFeBNKrHRMnMx4+26?3p?a@=lnEwns>jOi zY;s{5Nld%bw|pQdp@i76#933|_mzl^I^04gadUk98Tc3S8D-*PQMJ!t34;~W(Dz?Y zv9OmtTp;NqmNl^fZ%V}(nREgzWT(Ar7E%F>X@wEgTKf&bj8X1@jsD;>41>WsPnh_8 zR6WK8AwZ`4^?VbS6aq$p1)|wUN9lnMIT z>s9lAL?z6^{^@;9u}A(E=k1YCSx*z3E-Jq%DpvFu%`9#SKLe|F1l;BY>u;J4>?z|3 zx#GL!?V%6?f0VMKS~%;1bIIL@R~`q{MNu)(=ET?W%d^vYO?W zlE)KDMo%D<X?jxK zR$~ms%k9q_{(rFKWk2S){ay9mUWBGK;X^xO`i(cX$C`y*ujcD-ZRb`HwO)X^v`5}h z3BtTod?Ksqc9M}PR+m7G0X+Yss-2pwrtOm*;%WbXzUdA{pgIBFe08F3VZ1Rr_0GWw z6FbI4y$`C#>|ZMfk1k3~y@GdofEAF=4i@k<9ttX{`X|6lxw^tu!gs!Eer}63v_!|{p+6?cjcaep@az?JcU ze9M+Eu07l6+uI9F(3B`1chko1j-AJP5Hpf8>(fFZ&uQ#DAd2o9U0hb*nWj$7ecHW8 zz)2FX$<{jE-^GB>URFwp8KAlZKpPGOg=UNH=Wad!08yxkx1!9S9$f>V-dsQEr|Tt? z6wKP>{VW@O%Ty+?P&pUo6Lj9vR*^>j^EbX+L||qi^3FUu7@FrjI1#r0l=<7uA(5SZ zr|SK{T0z#d#MG>J;{_8zq}KDzsO7B)xpc~6dcwqQzcOOp;so8t1?Q0A^}TsfqTeOy z!9bl}sLQ>$!pZvVyhO-OW3h;Q+75Z4TI2dsvr!7YiQ$Qg7?^02kST&NHNhpi4QmuH zxF+VwNT)CnAg>9jC^^L#^-AYOaOvCpQx>UZ<2GlCHrQF%^K1G#axVcU#*x6nqK)`N zpfrbn_IfreotKld#NjZEk?|a$#qiMMH`%c)ELWypMd(&s?s)0j=p`+KN-PNd zh_QPrtnyV`VGW%KwE&DHFprfC*nF#@{7uCidI-Xdo#p(Si)*!pcLIwg_p-?dsTuB}YTS2Q}+_uYTic zJN%qB51&Ra&g?zfyFf1bfNCWWL5Ip+FbX6u+0M`XE29!O8LQ|0kXPf zW06SmBNd29h%b_%l=uxCi8YjaFfI7>+xm4rHeoTFX=Mu8u{5;v5-+)HQ$(5Ytz>@n z#@9qVY@A2vMt%PM;nn)A4aMzXS!>Kn!6memLqh0_v9^u5)R#LH%1!=*pCkTTFbvY2 zveba)`m#N__5AB6ZzT)I+B>YtOG@vUM(G6S4-`x4mZTy+r^x3DKPULnA~-7}lM>Q_ z$6i+Rne%V3c3F!X>-~v-?5zt4&JRSe3-*;7&k{~%E7XUErCzM1?-~vmGQqgAq%mwn zc*@-jJUC0;?3o_!3QqXxn^P64k##=W<~=WS{gkKHQp8jDCa^tQhGS^0AtyT9WaBc& z)P19RgshI7<}-ZsyW}v&*d;0WJBsTB0i0dEtT*(N>~-RbM$zWWVpbxc_ohqE4&y~P zSUG6k?e|PYFO{C5fc3|!ax5a@4q!K*anydCNdS^CH!72UuhlDO#hAQ0X}X0Md}zzQ z_9r&)7UL|H8h51!w$+%l@rbL+d>Pj1SZ;9e6M}o-+mjv|R)|^b2YkzNRb>T3PGMwf zRB^yQ|2_1$O-)2-XPI$vZ8{AI>f{X@7C-b%*J)SYQTBv$?(^kZRxX@|05Da;BlP5G zhJLw%<^(hVZ?xLbx^3OJynJo6g1l*(1f?p-JnVZU5%PfMpWpcS{fn%W8@aC6u@F}Y z&eN(t^!!^Ir!X)mva=vm9i9d=-+msJe|yS*OJ{tUI@IFly?364eWLHp@h(pn##72j z_z%A4ZquMUj8ToVd~}PCt{F}%z4FxQSVS32Kh1mdfO=`aw&XFk+BV?U_=sC~${yB+j6?V5XUlNK;#uqiZ*lB0v1bJ$zi zmT?FP?My3VWcs9Bd~m^;JdWIb2E;O1u?vwb>7?A@UE!H;cKK7>McBJ9-)R~Bss;t|o0{{)h9y;TKRIuKB+ zz`FsKi1nSzv;tt5;DcJc4;Z_-htG_^sG6Im_4tQlese0Gqi{GwJ}r2Yzp_?)dh~f` zY@;nN+o|E3E?%q>qy{wSA^#zv(ur;q>vR7?@^FY`E?KcKYk!H)ZouE+0D=L)10oXc zW0`_|yEoli$Xr5AqL5Uc;d*Cbzec8HH~22@f3|zNh9oLbNnZC&gwkiEc$D`7Z3xgy z5RxfYl8Cxx0T$ZQFlA(#o|2rNqV|6`?p8e(@ml#=x5SN$5=SdVPmqonjaGDV75fd^E4vqZ*xQa8?{4nd&rHBbadMz22ahM%oZ0=UZ#LN z(IW;j+|A(^{H?auxR~v!HR2>_Z3mN3McU2x-!&7rYO+v|Uo)Y#F3K|MTR* zyE)mcykA_Fk8UX3N2ztTq4mm8te`@Eo@ozdiyJc;Tk~>~LX??)6o!sWN%W6DHp0bD znX_u+8!9ufY$)eNkzCJ8vBxUmoSN~&Zqt5i8ao!U6*e#RY6HI5pk7s7h98!&-PA7m zob$zJpH%7eU}TpMRuO_v9UPtgiaY64B`xs+_i=r$e!kM8lc z3K9S3lf#sp*zdphbs+rhvYDAw><3lU>B5sZnUD#}pC5hXpQhf;Y8gvlKI#x`+Am7* zAosr)r7)k668fUNwX;<{&5}0t5=C;vP3#O*O zt%9DOnDEY@)q8JnTPDt`=o*gfMWbR?P(LPNf>pF{>9}I@M9^nXVv^)+*LhO!2Sh%2 zHecS#G9Fa*JHU&XxrWHMlFk`>L|^iV?Y>RAtsv_m$Vl#CyeNzblzU4`XolKB`ml^O zQVCFJ8B7@5sFy%u+k0=z(}^y8Z9rc90+v76C|Oa#wz#e;75;8>r462>MYzv%Y@!?$ z4zUOhAQdXiSbbVpM)#v6T!)ha)qd19pb_(%O-@{OmDTsLmor21kFxrTrQNIu#Di&k zQHuLzu%*7r$8t;ZVNG^JeoRKX zNk~RWdTKi|jIy1IhcA3}bpYsIXxu+U>Fd2`J5{}WV36ZxF2zx46ZBLLT8`OFj4@5= z8|7lR=vu1w%;A_G-|if1ytd0bOtRKytu!c+xmh_knO|MvQTRG=diCyrE6ECEEK`V* zo+|92^JZm^!s%7huVsY-OC%~(}hXlOCLAMnM+fAN(uCQvrb?+SH*o5s}mqeA?d-b-g&&?eZSY4YFn1PFKTZ! z6pV0j7)+2`%!JKQpbI9;=ZF~pev^iGc3&)Ubgvd1AAya>@)3T zZSeN%iJPvHLMfF7nA^Sd=^@VA7lQE;T2s&#?fDSowOfUA^m7{PaQr0^Sav3>j+%_n;l{Ix z@$&?1W!sG=u{ttDTwf?QJ0f4>qwg#7aDx^TbBu9=w~@673__f4;AR@C$$j9Fhhe#*>@s3wr2YhaZ=& z9yl0D#I1|_vK3@C97=3*?T4Po-2P;LMx z-C>CdKOVx>JpOplXYSjcnBLQ*KiW;!Vk?t(=WP!mJpo@PlB zEUw}rW}NKpeCd)`>-+D^B#9jn;r*IWQ_@l8%-fV}_0&vKX!j%f_bAE@&+EYou{#Pzq1ubcvLG^f3N zRvUcw*WmhjeBGt8fqKYx_utco9o0TijKxW;4Ts1f2Wsv(P@LMrGEBfTXrK~K?$&$l zKZpTvkLStM0swebijm5G6eC#B?yI8IZCK7|g_u3&_9by^>$I`^JCq?V;U`6hm<_7d ze!$eljV%diaiMc|RRF?VJv%4^u}OrA6=5Z_8Qc7Fn<_TcC;`oLdK={IA|12v5lP zwxQrJ;4BfN@AwVxZCydBtJ&>KR-A6nI6&lqF8Pl_oRdhpG^nmOHRm0qSnwKkDk(j@_*vb^A!0u zpML2sdxMi5bo08r$(nb|kBRV)Lri|+J)W0AT^^}j$GfFLAg^e$=_=An$s#`dI0$6s zid{O+%me~ey+SbgScg8R{m}^$7w(z$)jE&=xnVV{5Lg$)A;ObQzLh>Jr4hj@42s2e z8?Bo4-{Jy+YV$OXrVnz)F>$9LsBru~6%5jS^U~Av&j!?B*hy+p4`PXc+*wW5T6`$C z(lNqw5PAYcztt@>%^L47&!QoMyEZ?h+Bt~uzhdS+MdbGO^~jRZ}9u)!l|0m#!HM2!@@lQKrtp{KG_W?zc=)tV@d$f4 zGA(nFQ?l@zCI+OMk@za85SReV#P1mG;JidQTy41T;fq|$pmK*pd-WIJv&Wxw+D-2U ziF$H?e*5;~(~nQV4SBP+zFX55OH@)06!S@r;DDwkRCpLC6X5O@yr>{_%B{e!(lqf0 zKuv56P~glmKP`*?l5Xk#f$ianTe-kkK6&K9;y!8ju>LQt|64U`zf9xP7fYfTIE*Go zy=z6i=BxtswzaQsIsDCy=HI##ZJQ4rT)DqkdTkysGu6(EH~6tmX@`DvwkiJ~wxc_p zt6~Q7a{8Y(coP;uv88LzU6$p@3@%DC7Wso7pbpro!{6;15D_5EaZWDAQL@XTT{x`sD`0A|22|LN?=UtMmYlc#&d<>9$(=f zY>6hv1($W#$M%=0Xxu@slHsC3Dth?%AO{k1gwN^ROW_cE1v9{FjuNO?Z?n+v8AOv{ z!Fy0cAi==u>k_&NU6H)l+XRg;r*z!QA3S0g4GLt;BvO_4VL*s>X-hSd7;n>V6n2Po zH>)A5xSXwJ>Sk}R=D7j5neZ;}wxx#LDg@Q^{$N`KnNgWIvhvj)Q=xCvRhZ#4Z4tlT zRu{Nq^znlSRncK&NQ1XQ&qf_wD;{G=|C#e~>7{<-Mm#!>m-)jkRXNd=x!o@saLDAU zS%cSsRr1TC!P5x^tN{~uCsOi1{Rp=CG&e!PIc}BGP3W^_2D4s>Dr{g@Q1(hke|h6Q zOV`f3Ga7l5vKb0(_6jCU#dHG}siaK$bGRtw4h5ZZPRU@8V1>Dj3s@k|ldbOw0tG1D z4kq&1>S4h=`9O>EW4p~l`Z&USv$t1K>XbNj>jfVDb$gICWpB-7pw7V;+Rz-X0e~d^ zQa;;b5oE#4gDQp)?{%z7nMYlV)XBpw&c?xteaq8vk>-Z*YY~U9t}5RFRcV3aBC8`b zAg_Yqlkn!zM-=qa++}G4r0Qc+(n>FqeBv(LoU{^i5a=FYkDSgWS`<-^K~^tGt*oMC z-LiNM@B{bgY){0Uj7?Ix93yr7P=XKN!0471NS?}O)h6#TmEU@Rtn|Da=?U4#9{3s~ z5pW8(`*nKJ`=A1?pxsO$VV*pV-0L!z6wsf!eaZ-hf~kP&N^JbyA(3skIyV0NeX1Z! zv5m$|Wn8359Lox1hgL%(hKi8$8z0?|DqyXd-Z6Yk9gdjdWJUEPrXI|6 z1ml@ce&S@Kprr(Oshb0-+Xu0&-8+6_AREpojy{x$#m9Wpqv=9V?u@V>i%wDg>}XWJ zS+(0dfcC!$rZ%&T-K`1_ZS#za%sootrqTF0@tdyDFAa%fd$+#u=$MR1w9rKtgog?n zSd?w>l=He?-G>(L#3$2B3e=~Z4sos`w++laq969IH4DB2Z*t%Hej#8>quv-If-4c4 z*o09J5}QGw&t~$1Gjp7*$V?HL+p`*B$TOh!L$g`coKcq#j62Ej(0pD zk%a`46wq&TUz%|~ibZn`amN5Cxny+0fO(3(6+a8XY0&S4HfKYZLG&=oCHh(rdp8$xG18(om zTJ_Gwp8y5cuFaJu-n-V5!JWreo$Xc@p*M7ni2w2P;jTRqcZM#OhelgkLuBTfgQXIL z)-YMtlH~1DWj=|gyL(5ElZ|$zr&vwUPI0n;?G-1hr@?9L(#PzD9$CfFT=Wzuq~8g$ z${H7$`tHR(-mb;>sW9P~tpX<-NlozgE;ZGMDhKZkKxXk8os`dwG}gan{Pr*|!k5ar z^&~bGznMci9dc|iaw5A?T43n7tD;q3;sb`!0t7o3$d2RzdG{!MI2kCS{AJ;mPLNm; zH17}(y6I{LKCX$rYUT|mn`hH(|MwmD%TERGp^B(3jWs1~-_^GxzP!*eEyGr#P~Xvp z5Hrn`xPd_VPC=A8TGGAt=9Mw7^Z5sN3`l3xH=K`Bh%Ir+QyAL1>u2Lms^?$Wy>MS? z+2P~z%Fx`mKe{zKMU6`L61JWP2if;_X>>lUso*u5?Vyg-NnKUWbOgu3!}Inek4nq{ z(?A>S;;}cSYEl9Rx{0B)S)6QsIn2=eSPq`AMM>8BMhAg{Vpg6HxaW`wwF3`!vI^OQ z;)!A#+EDd6xOhlO9vkK~*V6}*>}*5Q8jO3_3O;ISzi1fBd9YR9G3j_U(7-sVwfisy zJLt1#A_6*fS1|96LKU7fGOjib(*f6Be)ifz-r`1&DvMi4U(zT}8h&vG9EGHO+DFgR&YXDE%nm`I?f2 zgYke0^Rxu>lIHYS2osT2qNB+J&i9O`x|Y5v7XT1YIn=9?R5O&L85bKrH(7DSCe|)5 zz&`qjRRDD+k4SVUt`rk?lnKFoU#7C&X=}5JlynW9o=!9Jd=M-*FcXz&o4;49v%>|Z z7`PEcY7Y`84d$?FkI6!D`D+9iwbG1La= zGbk=nW9D_<;`IjSdtC6h@4B+G^o{i;TX)=K#LsWC};j6TN~2oFO(H!^FLw33qmLYW0|X3h`htWq6K zjRa@cgBwy`og}GfpI6wlJREpK0z!Z|ez>0ep_tsXRC&agsl|XQh5J%_bRP}-T)z=0Rmk)_TV}v8^2i5A&gKg;&(PXa0|;*Fyi!#&;6Tr zsDNHGu3m|jVbYJpj}@S#L7y+gMP3*;s{K6IXTNTA0wpQLOYX;IsSEy?t4NGum{8$l zyFk){&bI>XIGqb{AGBq)O&~4rb%~2aJ6Qv`%dL-toszsSXxB(zbMDK86(<|Lq|k&v zOT0YLi*tocJTD3ZEab$-_JEPwqC!5?8wtR}L~v*U$`~d6Bjw3(A#3s!!Y-Sv?4)#{ z<;Ja%4P#m@)DjdpC|SN$;OBK_bf##{ci>U{&7j@O(fVlkCmHsg)Qc2@dn?-!npN2o zh`~qD6;*{K6&6SRxX3oLzeiPO91n}=>YekM!K7$Up``}ik;sIDjeBR3rGUMVLdSAl zW2%twQU0ptG)Vyu&1#;t#B|WJzsPek?P6I4IihaXd-`I!(;V8Rv7i>1ny7R{UUOlr1iw1ZSoaahB$!|hcWp2L~1}R)_Q06kU!DM-Mty> z&$dgPSAMX(gpM}(m<&d?UpeB@I0|b}&$O2t-lTYG?ZJZVTV)_iJJYh3xly`~{9(*$ z0(Q$NA|W>i8)>&I<|~`a7R>~IU_9Y_`}2-p9`5Cbe`=be&!1npRz#x-SU=DUI6{zy zYO=HG?h0xmyB>DJcidkvPg$(u#|T_SuS8`)=0-XMyYDIyG})CIsR8s5&$<0w0eywk zp|ZYCYhf32d3w1b?)VDc(<)=~6eBFcfw1830dZ519Xev4hUXroQ&59-=|UiDfnu|z zVaIpui!~rSuy8cGiqr&Ot3aVr-Y>$xV?4RrXnj$Vq(GIyU!g#~ZXoNcdwIVPpqIFh z%kSBq0=HHG+93H!0;GUCnMih5@vGpLlBrqP*uLQzJ9>|~WH+Q^-D{U-;w)%vOS<)3$>yeG8Q!tt2vdgI7T=qDl* zOWTqKfI@{Zz%Z7iH*5;kz%hn2JW2!NSk!xdf(uyS<8$^q_96|%{MCm|p}^lwzO+-^ z6&j~lY=D+p1dwHkVDWVC^9Mq`jR{L(dSw$7E?VD{ZZ9r*>-cWzUvBw{<=&Azb`$AX zldh&JpNp0fk~~#?`1YUS%%V_RDlKVa)c*WN`4aGa^;TTudWr+nDlHlOem5!&=!OQs zqAB&9(=9yN3IuM}7 zbfYyPZ$SIPZk&EO_j7VS+pS{Kw!1?l{X8v4gKV?B0vLT0N!_#CH1L2ipDb8IKL$ck zUkE8YS&*h31_uRy0+g>o=|M`;(NLhTDK~Fc-37tIXE(uA#kVt3YLdDPUu8;fD>j~9 zHP4QV1g?;hXnEDvAIfB!Uo{-vLS8%S!kYvP^LW~%jSB@Va>FT`ZNZMuyq4MJfb8eS z#`q#Bw7egM+r2!h9{4_NHK{)CWIjn+z%d>LoI%Z3Pak!CE#v$5_Gp>iyrfF2-qzD^ zT}kN4rsrg=qhOv^Ia{?(Z5+^~9jo7yOTf1lSq+BPGxionah}0Q_ML70!<_(n$NHzY z(7kwfYMinGmK`3x=0o{5E?%zqaw{>WwBQ>?KbWxcrTxE z`C2!ayg!MrULW_QrNw(&a)e^G$9?MC`4g&Fl^iSRr3Ah#VzsgK^G(I~9d;)6Nff>^ zn`DeVC!1^-O>Ve~BZbx%%xI?g;3(#V2~IY`Kioy-fbES;9+~F( z2G6DX0rI;I0$o2~_t|MYSD$OR(e`Yt?T#BHqj!^)$4{(=(`tjCcbVKANS-?Z=|(Q3 zuC((?=;pwYxUr9jh>tmJ)hc%s9~`s%MK#G5_eaW5Hzf~Nh5LRpT~a_>Yi8~&U}{e+ zl|Zn1sz!FvHHUt5uCw)Cn(`Q3El|23X}Okz_M{2x-ORN%%%ZdQWR<5=Afc>YY?(;g z!L7Ged(SmG5kQNi87PyUi5K)C@)v6W`bWjh$#0TCtnGJg-tY|&AEZ@dfLXuQ5VX_U zR`K`t7YXDg@vwvf+nDI^#`kz$auPZ?qbEfaLv6PvNt5EUe(eJ2U1JJW2N&e-ssL!$ zUc4kr=L^f|5D%0`F2Z|RH z+5YFx=f^{yZg)Gsoa9Kx=`?Q4iGyt4m2ch^JeI$O4n7UyK&==7TLFcbhy&3J;H%=- zPawh#j|$pmukYtmpDCu!HJkw%+r>r3b^o#SMs{?rA^0q2S652Z^I7jvYdc1Y?*_jD z3mSN_L|;BadCXFdx-==8_@(CU8nZ!KYd4;mzP)z8E7YS6;q9Ny!z&2Rm;2K9Z?jC`hfxMfX44wmq~8B*yL5K7XFByAlYc=5b$8CAK_e_`Ax;dC z{n3iRDM9IVOv|k&M=G#B^r-W(uuD|b8V*E6Q{riukmVQ1=>ZIGpvFTrLp}4mKSUGXH z)VNafl~Q|yKP*AkMR&>Ua^|y!PfHXa9HkEX1_6h26^ixP?k~_%kCkd7g#sL||M7OA zs3GbLSA>3|&tm;VlW!X$ieE%9y9GuK>W>O_c| zD&FS*V;k(J?1p!_EDHmHQztb);4rJjxHmVRVDbe+bkR0XdEfe7lj%gYGbv8Yxb-pU z4dCg28?fl&Qm|CPjlRhluXka;Pt=mTQc8fcuFnboF+!S5(vN&x*65zOdgdA9;T59EmcQKq09Rkuyeq-lITMKWclCx^$_S^>UkgfDaw$S(J()glc`_njo6j zXY?3H_-^%H=GW=6@q)wtp6NkLW9j~lM4S6q`qj|gMy&q2Np>^1aMimqV*Q{>1|6wO>admkmg~yvHT@2+aLtIDp#B{_UF!or#~ERqeCxj1eLb%Z}lm5 zw&)fc6^tDNTI3IVO@TMxAlWjT*;>`)eunS5a5;?a3x@vM!zRL|j{p1qCkpjjk#3oS z2HS)R56JcynWzT8n>+&YyP1(iv z{SgVVOR&Qg`irFYYP(LO`SLRtSMLBP97NwB>zVVhZ&j%Aa{s72^DDp69?&TK?)@$x zSZSg3#7+GJhA~8lfhjjPet=Wf${Eg@CdN$fqlBr{YQ2J^6CZT z<8bu6gZS$?3~15?%1ul71E;avP+DWNd+U2dwuQ1XtyxIB{7*F7hV0epH=~XYj!?7P zv^!cW&knm)pyt`}&{U-{`L*i$(aN8Lft?F$zbukw;y3lt=eS_r5t<{0@$x?bO#jd; ziNx``URLSPQ+?rN+cD-m5W~QS2C{6I%glZglAC=aV3Hf>2*>qb-2RKkpG_}YFHt8711NG$8_U`Z>XK%BOsaKg63d{hL%rVY?5|3M zHm@ZlZqS~Dk_TG9(t(xa@wW%4ds0xWFoyIq9)_zorK;OOP*AGpy$kL|GZD~iazji# z-0<=*4{2p&g%}$DsmosUaxwhx$a9_ ztBI1dm*GYg$=$LcGh<;KjHT{A4&{eo8P4oX%rvA7Z)Igek?Zxam<87+W^LQ7yn!xh z%pzu3UZx++gQS3cXl}Le1GwV4;=2PSQa`Q`Rmj=Yi1{c1{M_r?TV4K?h=-Yv1bdI= zKv(36d2}aNI9AYNjsugv4q@s%d`RKn3nxdJe=suOd3F!#iHlRoZ1v0lmtxF0PR(SW zIuE^ujtfd{m6QVtF|QkO6DV;KG_doj<13bv%StXb@ac39>WNYt>{!Ltle=~Yt1ev3 zua|wC4;MAFMd8x1Yp&_*RFV}}Y!&PPt}8jP5fP|F`m7P^vnww>uq&cdOWo=9-|v3g z(g+HXle4&?oc0Cu{chL->74pVkFW!TuosRb(T$d{hHOSN$c!v+nwsc7hf!|w8UR-w zeu$>uwA(GxXVV*3o@G2QneYH=L(fpJyA!a}=bFVFz8pZHJ83CBg^YuCf9v}CavPf> z)9}j4cU1f|Jt+1o@Uo6L&mSM@P73!UBLan`bpy?Z$4cjmjRw>ff2fgWKUkqn!_DnJGC z`iDdW9Vk{;{sGRb*33$)fuI_3I^dOd*6z$<6#&DSs3+ysRK%V1ML`0U@}6R~**8Hg zeH^b7iRyk>v7wKtI01Ta5qSn@`lZM!_4XBfM5zyIt&H`IQ+l;Z^I0nQ^Z~k@fytK; z#%wV%ZVuM`i7d*7$a01}dt~#(5rDo-&93wy-UC%B zuN+GiYvk|zyJq-MHcq#n^emX$1)s+4D=mmmqIpG4C4pPxx)FcFUxDWRb*^~Y*;T#r zevapz!0)uc<-dxt4I`U>0&sy>#g8)78*D&-mH^zl11^T)M|NKSSh;0dA(xux-0PDL zcwr#`{=ZD?7H}fCJ{qE? zQ2kG=T#-DP8_+v0b#%4;reKN8GoRj)j);7EUv{{NG|ks&Yr8`L9T1}_!Spn0Hyr?J z+;mSwWuz#OJpwIrUAF+EgYLvZjjd5L4X)(3w|__SUD_}nSg zSseAhsEO?Pgw;+xczX((@{fiXlWpcIlaf0a=K(a-v^oAJqxSR-Ko5f`@k~>k`99{o zsb<7}&FBA~8RQR~D9TkFj({M&{Q+#YUwqXeQ#>3Az=HHkLji=$7sa7^>D{Rg;xi^q zBa09BYhwRSVzA~Gm^5JL7ERWdgEfcIsh8-gwA(Jf@;k>M1enu6ldlcjksoiHxUlJv z7SoU{PLR8RzVT0Fs0U8F4^=HXjy$nVM&g+59Wsj3x#96$K-4?MN1M%Kk#^>VrZhgY zTsF!!{@!={A7irz%u8lC5g*FhB=9@+GRWyZ)kk;g#S!C5%l~)An@udf75*(P7n|dE z{rwBxS;x=;)-V`)B(O;N2_We(U-;o(paa)`A<S>Ij8fI2O=1waRT z>q)g`xA*{!%j5rAR4M#j|Cu7sK3Y$G6L=*CD(e}!U!!!AK7 z$1NAm*kZi&qD3^qw&Dj(G#=6W-LzxK^g>G&cJxtl!_}rBd#Q z#*=Nsc#%M}ygU;pW%ANOVmT$%dT*t>0k>UJs+h08UZ@G50?^3Cp-FMS?(`^GLMfm+ zaCLE#d{hgf*X2@?#1Gyx#B7S4Ll!N=DkWXYrjGXRWLxQyG@6jKyEwqRBk^_O)$>yA ze7jcesbVA`1r*%|U=xw0K2W{)8_IoKlB_VG*cN=FzPuMxC=-R>_yi3~1BX20pPzZy z4eI)C70nZrJxueit}b~FL*(J)fJ)8n{ye(l{nVIOJI}6J z3G}jIHWbXa%n40duCrJ%%6{<&e?e(HjEJIOvES(k+xfs~;=P!5@f5TA6~ws!7cnggqo_oVvb0gll{} zUmAI}$w*)t|Nc3PTO0qdR^=2w{Ej**|Lr)9XAPQKLPd_$0+a1u1RB2OpMz*e6@|+t z@npqjR0n+!`cq~2;J4qiDOo`{f0vv0RvL&D0x43Wua+`KGqWr`^D%X?R&rXnJOOV; za;}esO-Py9n({dXtQ6lj)rAaH2T<^g)d9P24?v+KsMsn|)_P`*NaiFb9=mNIXWS=U z6cepic8_sS74Soj?*~3?4Rrnx2YB73VRu>C$69@gnI0RT4L)8(f2IMgpA%=;pzAe* zBg-&y$8(xPq(7_{XfD8pfx@74BVtrd-F?15om3rd|pDvD)>2L4ve* z?E-T0WFGi?=xm>2>AAmR03)ys8v{(XCjd%rV+GrIE=5De49R$jKC6}r-;x2ggTQDh z53#1}AA*9S#n0JnIb4<+`IP|IC)U1J*iL50w0GWiPLn8V5m0eA2~N8YyZ6hTNqOL! z5!dV4V>HRv{Vn-_Ar?CYD;2{a%(?v^nto@U=VfS7MXTZaw=iKbAG!GoqI2Thzc?0Y zEqqZH5BON%-;&VU{u_Q36Yg9a^CDof9clrubtz6fik@pTz-^noLMt7vvn&tHhHmc_ zVOO?>qv!ucRHn5OI2{FK;H91VrOfjugvV`25q-mFvwo;Rd2ea#yle5nl7YDVzgd+8 z;7X=2Si;@xNN2%!rqS{$?mPdIKt#@X{jWeMdbItBoz!1aeNI-5mePd>ZWI%1Vma}@ z3Ri_dXEl@xkg8^SeC%M-u~^WK-i17wd-urB&c~DrO}gG9$0RvLIrLOmagLfEPAPO_|te)a(Y~{|2?t( z5O`TjY*Fb5#%M2+5V$`{Yy+=?^Fmv9?O;#<5H|`1O^5JWFhS?*GHv zdk0e4|NrAh>S&lz85vPY5|xp8WEEvZrHCVLw@t+%^Kes1O2djcHf5I)vWgVhGeRZv z*xQ-E*L5A-y6^XWzu%wlA3uK;*LA)2b3Pu=*8s4CCgN@6j}T%6z!V^bQihZlxX3J= zXieFz{j4GD(J^4Q-PUh@xS^?@fxsaXP%S73 z-IqYe&?s)zbIpGz!0l$W3fvvC;l-PJra`d;d1UwQx>H#spo4^*;~GSfuTlwDmr!P2 zM+U!nsVb~(FtJxaRcDGs7*wj!v)3hAHYE4(%(Wi2>1KEDD|K&qLTf)G+O+)myDaIP z6<2g_<@OQ}l@M_Nq#X2WZf7obGSpZ;#Zqz{xS%d}hPu}BB;NpQcTcf;gYjC1%%yQ5 zUGb?CoViI8x!^Dr^%xBA@;q+lJyuiWrP^}Stu}4>&dFyQH`~qcgP!aXLF;VU&8(xQ zS=W!lH7NI~K{n*X<&mJR4o^Wnsn1`0V+{8yH!i%O^7nz$M%v1y7YW`IvKeM#*dHoM zbzXIIZ_u*^JbJYWG1uZeyo;ej)m!P&58dD(3IikM!=6K3=cmyRQ*5v?vsR;&OrtxM zfi?ktTUCOX5TDPo#25yjsUwi^9N6O*Vs+>wgcvo>30=C2rOIGSe6CV$YcIduSM9_j zVK~l!7)rMs&i}!al~n!{7zNk7O(Sl~R_pkLi>DwWMX-B=Fy- z@w)by;H(t{c@Vg>Gc|-CmhAB9`kkQD@O^H25+JjPeks= zDh8p?k;nr|&&+#nsI&V%nf4fAhpKU(AEFlvDdCO_J*6v27$124KQdd5nLkgYq*88DSyYKngT!&ja?7m#nM7|AQ=^5cUg56{gJEo6A4At}B zO;`&oNMoFWP^o*(^?pQ)EE0yL=ps*+H_}zSbs2q2K40}wT_+=HfrbJdQ)(bKeWkrxoo6SavWob?De}hL*xW|Y*kQ?A3NaSc+!Nc z)vU^iOu;QbebN{sV!Ya?(e*x+6w6oRLKdYC#q+<)>-17)_uV#2+`Zv5N1MMXd)$M8 zZwA_)jipoa5aO@SQj!Xhxvo+Y`BBEbR6vhT8~SpyAfuexZHz|uVj*gR7cZ*c6yii& zA1pG`@!zxfCOeH0JNFQ4TpC^{s_zPP2;JvsRLbc!*K49~AMyYRKU$91lJWujfzHfy zO+C74wvFnFyB~%CdA700BV%()#Ff*RV$+SaUlYR#VQkG_J>mNpgpQ0Z7z0!4*L7RW zR62A+%<>E+e?C>HB6PV?X5;wo$&o#tn=UQ`2uYzwM*8k?+pXHzIukd;B}{$*-)Vog z1xrkN%|hi?o}WzwvNKmOLgFIjq*}`3tNE%$V##u+hZtoKeMb);)kYkAhb`F9>4>QL zj(4M~A0h@dM+;iPU=h~W(Ed9qQjVRQ%{GySxoPX~7MHlu=RhD_azajRE2kLRY|=YP zeTMXPhnx%Ek&RxFFan`?3loINI^sQxZ^!yEg?`4yt~Sjex6uLGhENWZHPk`I!}SZ2qb5}ozu^@NT6X4 z_h;d`lt>WEABwMwtKv^u9(LHp>wISAQ@imWF-qA82@&|Iq-(O(KS2anoZB}?CrX`u zSC7}BYLr#tNTh15ZdTt;NlM11il0r8E9{(qZKXgLrMdk9B=rzOFY4o}zHk^XMjgx; z2$5>e4Gd*RM-+LSHN&+&Dd0j7P7Py>(sa4Q4ersNcL1Gq7Ag4Z9K||Q6E$*Kt@d7a z-PO_qIgQquJ&Xu4Cp+&e6V=i9P_mBq~QoBgPl9hA6 zw*UK|?UVTxJ#V|fViGc7fBi^D^Nn|@N{8Z6tvT*2c7Ob36qXylvT+l_t~s~pYJLIH zGji+u+!%-Rf$9`g+_oRT&fF(>=9+;e%|iIalK!bo4d!qdPtnA$XVhq`wYO{)9*tV?xKlQHMRz=wVi_Ch0J zH2DL(de-%W7D@T%7hg-b{-EeBe1vS!OCNT+1NS={sJX?)hgd)zd12{# zPp$PVZS&>4c0mAeP8V?5iv}(%x;L=OQzD)~P-%#k^8IaY(qEgIe3C~PX!M~x9)zKF zaG}7k6_C>Wz}(mF9c?v-hEkjm8oQlbQ%?QdJ#5RB*~3S~*bpH4tgP-zRVv@rhAGj> zpPI#krqrTfZc&SS y0@7!$xoFqlVEC%}!O9GbYr4VsYoT1m&TUxi3PagXs>Zr80 zeXiWOFs-fo{0FgQ9Hn=lI-A^W+QD1S6*`4cKEhU6l=T;zc`@7 zR8F{QUKbdiyKK4eCcx=u2D3r(PL+>(MZFKDhzeF`T=s@rcS8iAv9ebax5NwV6}eC5 z3@;DbOT1$_a_9%zn4T>N>o8N-L-rmdlzUpdYzdP-wN0zX6fx2m$sp9{JVgu-!Q}se zM3THpoI)HVRO%pDqTvuxt!A#FS9kkyZkvoVFCnCcnHJx<=^Y^SR6+z{^I156r+K>pCoY83H*JA=8Gf6o!k(^{{))R%0%olpyi?UIhr8NunYV)cK{46p&NAl9F4x63vD$NY zg2tu``w7@~+4rX%?i`!z*$zxta3XcTooDdeelIJSQ*h0<`&CGd#H^pil$Q%D9<`TO zCN|U55(7ESU-7oJgi1LVTRr~7e|>*wjI0$jY5c)@yoUkt9#U-37yc6N`%|WrZWL&< zK^%$&&j|)G^m9E(TG#(on0O3tkU@&tuN5-434JfzLDRbY^S@|aDAH$}@d$Q|#Srri z(EPHA2$vmoUg(I?ZdRz{40iBDm|~3;amC{Nluqxu=AT(1Q90oQ*Y*^%>vEJL1yG)9Wuf*@((nfGH z$Yr;xcxOo;fx<4}lY40Twj{krfBq`4Px|2sI&#nx8GJ(F!r10pKW7D->7ld+x z|A6tF82FuU@7G5^82Bc2*icLe)^7+Xb>QMGR*mF8T0+drrxS?tA8??3+hw{&#E11A zh_Qkni9<*s_#YL0AWm#n2BXH#rG!EF2Lxv!!7SkB!Q`+slQWPZ^yOadAECIbrNRkY z&VVEZNIUu7cug1R!Kd@(O54%>gTjVW<7$?tphww#khO!rWrq6pp&rSfL0(kT@;0(4 zQ(T74r-zW$^Ln7}8{{5tsp+)*+bg_Y#EFX{J_I{`?wN_oH2UZ1qfS zL(vb;Ldkp3%E0gR%;$;qjl|f4_5(fs=NTesrkGVt#HqH*1A|qxHZ%+nS{ZjDxi4%(k?^Vc(7#fS=MI+Zgp}Wkby)vvOt^s8l_V_3Q`3h#vQTk)f z3$m+f5PC8z^wD46MFr8C|EG5=nv%$_V%6c^I_zU!o)YUr8ecvhNH|YBuAc82t!q)j z9><_STFEqiZOSWer%V*0w675N_VWeHy`eoUFJH!(;KM0Sge4G~Ciu65Wx z0;&M=upem3ckT;bi*q3t7_Oi)DYr}{Z_PA^;6TZT`6A#ZLCf&c5G;GJX**y2m&0Q7 z4?_or^&6g(;s&Ys-)7GiE%cL?EvOjOuY8=gbXa!g~Y{Y zd9oMLl2I}pZ$agW0dWPwgX?`|1Nhk_*Au8j5$f}iYS&L|sgnNL8{J7^Zx=f{4O^ui zeM^NX21g^%OIM9Gc70DZP97;!T!J`XuY!M|u6vLGX3x&DtRBA;9d45xz^bx!U%)sA z&+|8vs3OWTipU^iI8mGCFR%7nubEmei05T~mIpYG3xx?bFqsE}$fft(2PHm>NPu+~ z7xlY61iBa!5hgkR#UNTdt803-fxY}_w9;=E?w0i*d;fKksCgb`i))t7h^wtMU&)IegR z4EyzG?bgh*`1!E-HuUM4IXwP_ZhB*3=X7g#RfMGn8#FRj71xFx_p9$aVTZAo3q%u{ zCs`i@O`l*>kg@DOM!GLY+4>->*DkOA1aQIVbE90wJPTJIZ5WQ5?zke>>_L|2LV!^*0fFn&wyn z5lcG;XF!Hv{9U#%EYW|jg68NbP1ZC#6!%Z#SEq?*;r|~x4{`wOBJFSEToZ>tb}ARa zz$}pG2wq+^Th{?!z{$s4*OZ;XJWF2+!1>)=fgo={m;HSb{*#_L@#CMB3la(-Ly}G- z!F8Vqs>s7i`0x&WpPLJ@fx87PSLbfO?VcyIPt)C86J7}bl_?H{*#G#GbAP0~cCQ5l zEV?czcjXcI3ZPVZWI*LFDlQad#zS6Y5xbL;LrDKYV%`!O}A?44U zFq%ap6c*=k*$*i;S9;5&)a_-@R_S8Vwca8W*@C0Z1O0V{;RILD#Wz0B-g>v%XIu6# z&0F>=|B*84ER21YP_vtOwY!~tx0tX`)Q_bH3Cqq{NzUsZmmTV&cOWXZVV?xb71C5J zz(qzc9>Xtl_7*lgvG!G_XV;)->DbodK;Ah^D3+d?TacsVGRblFF7>cE6c*ww;>7Vt zLCFzIe@~3hii4qA*Egfbq#_hI<2nQ}0HvuGV@$I-wnzk9^BdHlV+2bGm2DB3ZXPcE zKAn|WPjZ;UU)h*oe99a`4ABvC)_p0R-HTmWfla%vveM{$ty09&(BF+ju++U zKyk*^jP1>Q?a;qCx1`8=ZJz#*s2XbDG4mIHNU~nzBKlSp`)$^KP1YekTofgKT^URq z#@`_8h6Vlnx8+*)KF8}rz14#|X??wdqEnhh`6^f6Ghc4LJS@!>IE{@ukL9QM%)j?t zsEHlAVObTJynU9Q&4T|?AabXDMP1u4v1VbolYcIii@ur$mepB)s@7ISGwMtc@6)C% z0Nha4p4iHWZzBT0Y-XZTf7gMZK2gJ|GF*W<4bQ95)n-%gYtnU>LL9h_w6<_Eag2*v z@I1=#qvxLxj-BAaFS14dP7Ijr!Kr*vGTe`yqyhM@zr*cMXb_`Gx=oz?7>~BxS?1iv zF)4hH-{Ui3>`biVC#F6||M7Tkm~u?`?|XHonesxE?HpNGa8W(`lXBIW zdte4nV45eWQxfKm_Lw!th(`TR3*L;?ZZW1n6$iCX03gIgsZvVnK53I?@7x6&qZlzy zsAy1#!xe}7^)%B179x-+GrQ_FEH(C(`Eg)My2X_Ig!H3txM2ovdn0l#dE4z}KStEI zZ$i!9{4<7X)=k%13qHEVy8BpCnjnb`1RCU4tPXI0u^_M^q!HhyMEQM|U)V+=#gB0B z&B5FkGY&!S9GCTxEeWbb_r?Wc9hjoJDm3%>uvqlpKa&*Vx7B1t07N!0b3Z{aBgL+a9KnHG+ohO0&<- z%zt%Cu`Via+kCo}G>eU#H57Ie?bZ-W#Ned_qnzfiX6p95nOvwUsEP11?TLu#nvF^& z?l#UY5HW%lMC%(gDQhBo{w%ShSX59_*MEBgvBS?$*f{3*^HwHvk`ntCyOxJ-1S(M- za`uV|?$Z58=3vFng44XWQ0OHxu?g`9lUkSa1OB+_vAU65rtpz6{@^AHb$r1>5BQVm zEy@iS^?=N?6a_$rLd0>v+jmu9DVOIttt>YbU~(|K{Fo#IQ`~)~D_IbMv39)kBV*gj zvj`pwT`WG!4y^S{^ar#V1UqKxMv5p4WBv>a3tx5#wzvwsb3zThm>7+=W4Vz?A zzJ|{}3}pdG3KkNUI@hm282|_vlkI=OkXqW}sOL*RSf)`b3jGirpk}4F7E5j=XP@b#MZ_}r~-Z>M3$G>b< zAf90Qe1EJD`!KqD-hHf2$uv=aYC1#;g0MS3dPwL@-s@B=?n8DGr|ILTXwNTo`eBzI z@DExL7!f03i2E>bpl$lb%f=MvVD zEF(y==t1%XHVqIK)k*UQK`6AgsZ+y3=$ssiBAJ{j!BD<>Slsa>}9r_d<05EB;PSScBI}@sd@wTSYtFtf6k3JY{UY z3um7uG$en@N4D7lGyA}_CYFaBmk3bkQu!H7YH24dW0a_|VS z0*AhMsV_MIV7UfPvpl4Gu4>V+oNr)?rPggRv7mC*v@z3ZX-4f009Eppe(3IO)(i-J z>XoomB(tUBmJKTNeoB~Q(9{bT*I2y|N&@i?^M)#%+J&c%lyILK3sG+kE*GcN7+Cha zlD1BD$0Og3HS&Cp?70C^Ivg+ZKiH+1h3?pd;LHV%4gKghUS^d;MjwDad%9Ib_~rZV zDYXk2Mzg%}Xm5=D_~8J8OPz8KuGfM)5lbPxf@W)XBRpL@gd&COx*D@839=H@jT}GFkxaf?T1eb@^>Z2Y` z$xof(+F>G=gvw8&?go~?)B|gQN}syQtfTtWb|0Yxiad)n8b6C?Qi|e3@Bx`&i@_2`s0rskjmkEfOiwrV+0v7LB&5XC@IeOL)KGMoFyq-*~ z%c46a`;i>uy3!mVRGeVLefjdzz>BD%mCkJH1;5gV9hUyP%*#!3M;Xwbzvy(~LOG^BeV7(`%%y>rl-pKyrLy}xaF3YgqfLSMRT%@l4-s}8< z4;}Hsc0o0)7~(;ZR!(sZAgbayFb(s=lr#sLaoiIbYZg&~Kxlw6-nZm<&V7U8Z8k&-gi4Yp-=48pn?Y&sUgZEF zNpVhBQ}Un~ww#TM-p> z0A<&h;(y&6@k0yX!o*vGAY;BK_q3$IHLO-Ve>Pxr^8c2zGK?kM&#w#RhHSGzKMuu9 zu@L%m?TcFwp%6*LMb0CeQ5oRHr-sz!z`~?C8>`IsV*q$^zJ!G^CF5X9E)<$ayMfkq zjX&s$9)qH0+=>M*+;4z38vqLq+maPs;7mB3J3V2I_!0m1jQ9q>DSJgLC`Dusg#QNS zgw~jdDF9b1q@H5HDNGcm+2iY{M zW=J2qNL1E{67-x|mVV=OL#P2gD-Z*zbAtXG&6U`!Kbl;Ciwz!g1odpNcW%h`b}$Ga zS55>6PQBGwbAU2ir&O(D6Z#)OPa-~70~UymKl4J;C_tr;-`@CO$v*Wtm=@ss0?-G1 z4C$e#mWI_AYu!u}*UWW|qp8+qv%U~I9f3G_Xn~Lgpn+M|5a?fi1e-dt!KM+CNwzM7 z?g>OlNI@V{sS(zn?e33%83g|7Sx}GpR8|!{fs9~B-m>Gnepxe$Vg;VuI4$_EQ3;l` z?mbD4?PUUA%y%&c5a7e&$GJBcHE!%;7Da|IApiqiFn6$3JL`7@1K^~1NlgTZ0s|z9 z^78G+wB-iTXF4|*9B+CCg}(-UWE35H`($*#;R#V1Y-> z%Dn^{HRHkc=g!D~eI|vyGgAsaryg?|ib*)mJ>(Fx1B2mYj_DYE<{6{GbOYizKI{fE zYauV5)_&02trD)ILM6eJs~}Ti#_2qjr_=luFFE${$*ZU7uAu6XeAj1*XIMulqN zzRb^<0AA(ap1aVZ(Hja0pA+JnLJxTdN4P(LSp$xfF&!dNo-qVw`}zLv$2731H8@z< z80oP#0kzr(fVgmBDK#nNW|$p>E^R6)ax+S`p2UpF=Rg5^m@dh3)Zrs8L0`vW?Y>0@a@Zr^19u#bGgpx<)(EsdDcl>VQ-4^HSw zoOGNLBA%KscMiOgFwce_^3Vny8ma#(pJ^buB)7oZ1!XB<9KI(8tf4^ZbS)y_NlLjC z=+PHL&_-G97WftIPM!3rBabLg0Pi3RVgGyV`D$Y&R1fSyCj03A{NR#KDdu zwr&TS;O6A#`-saJ*#i@T7x|hXh_p_zs}V%%xVi?B{>$XbgYot~y5wSu5Y(oEWmFzd z3C%0Hei6JoqWNFiwF9U4watMAy1?NUG3*^WNoTPG6UQ%3!l?L;|I)CVE7%Wmitd&O z70uTE$USsTpEE^lmcG9PZhDh{R?MaTD}i#IYow09%=8Dzrbn?H;Ora+EEG2z8KWUn zXct*0!wZ|xAhkw9F-lFI8{`DBO&IUvnx6L!v!Z)+jGnb|NF9;q=Bd@446Uto>VJ7L z5!j0-Y%0S5{Z1RJ4KYw`3XO(%q2rhH`oDBmFUs9YOA;}Y`*y1@#d><*gn1o!Z9MD0 zG^rj1h(+c!wGi8IAHk{7-5%bIvxDAES$%gQehy@S6ny~T?(ST{fzs0J1icwccm3iC zm(13dm77p14CXJFxa3`aewLkPEL?PFL%eThxtL<&$OX1YugmJrRC0mtZFt!4QWO2& zz2`}v=}GUjRU%A>&Ax`g9s~G*y^`Ij-R}kkk}rk~XN7`eDFi-9eC%3kIgj=Ps#mfz zzjaK0$N6DG(=`F}S1$>g?r|PJm!aN7&qBATVya4k{mU2;V_Vg4_-uIv`1dsznh7Tu zCqltynk(w28}B5G*vMqu)P__6H@@2j;x##Yk4Uf{yK<5I<_N=ns4HK!bN#lWCIBK_ zPRgN{tIjiDj?^Ja{R9|Fb@fHHbbB?rnIeib;ctvW-zZCS?mcLeUxQn2^RGibO!;EL zba-SD5H^lPbUG<_y;^{x5%pN8h;!~BS)tC($(vlOkf~ZO^F5+j`U^=Z;rKjWIP+#X z^&HgtJ)|H9bkWKYe(dO{`|XlYR_#(a85C4#pNFm(t)16yTmi+Rhh=MOLSoAEj_cyt zGCOrU<)B)(wx;BKwZa0ig>>$D?&22J`EH-sWvINfhkJi3F?VnrBWZ2X8oDDQ>r!Lq zM3!NA8b$xM#!BmjMg+4ieT7X##?346$hd%`%YQD%W&tkjFA9P(B~uK7z*}dO*7awT zj+}*(_DHxTS&b)MbFmp%&?}{?d2f2o8wY%XN50j zoNV?!W!PHQvjMP!UPf8PV3{P+*q!F=V?46ZSw8J|nr9a+~Gu&dV8$1{@lvz6C6r3v%Vko;Vim+{76&78&jA4{0HLsSdwD8u03L z`9d#Ys=4h(;Lu_sjON?nrJA7PU487El_ffM|5L{6^%xG>gM@M6mIEoNgU|$%I8W%i#@` zF3%fDcf|*mqxl1y;_ix}`ETh}#U>7p8QMuc7@MN5h*R)kep~s;N@~TWV(l`%?^m?6 zx_6~tg+*zm7H43W>w+KhfJ^dwxYWgX!~#8=Q7T|Q2Vt2Mtr+Rq!t*9OfF`1g)p}8j zs_iE%l$Ra>%B4gk8Y!6J%sb=1Wz6ersWW;QXJ7@w0dlkY=)Os!jSml5!Qp;CAMrJ# zLH{LeKQwT@jJIRsV>>cC%Ogw@2t-3_uW(5DvM$yABmabu&ot+e&OjdX7&y1tFQoZGE*HXX-@oa#O`E?~p=C zKFNQmE$u?Fo-eOe$!>Q6s}Qz3N*II_v*_6t$loDWqX#Rq#&Xi2(j(j-yvxUh0d1;p zpW@?$&hsvy1b`VVHM{P~bvit`*fOHA)pqbt+ExFn$`=_*+|LF9#Fz zExb>^GX1s>;)$b+RxHv@ITNx$a#PYKjjxAoS(WJ#`qT#uRHFrVkQyfY_$p@$MWs|S zqI5~#JmYC|jm_T81-2#j#Yn-i!6wNvzKhw>Nn0irjE)>l>Vf+z8{z2mSy%M|#1%OM zr}x~f3Eo>v!)YGHNMj;a4<7hZ?(M*X3}Z3{!7M* z9mBjLu_=~_ggz8SG()P)UZEgB&>%qzh z_Y1qR>B7FJqAJAis$`>0}C9-GYG7?O48B{5Gr^CvcDv5;+|mh&EE>)SkWO5@48CV0Fco8 zAq(DOm0pdPL#W?N#0zPoSO^6}s6b94eid@tM8XwhYMa=hod&m{Zoo9c1rp`>Kbr z`Oi}7K@waO$dYorcligJOu|**U0gvA-kp|>kWtrJrv{pb95{h zGwVqxmlV$iw5ZFw%OGSh?i|nr)xYmCVjY z_R<0eJe=@bdc=38+4T@yWa9HzKlx0~3OfA!3l2Mbqq2xshGleoQcbH4&2Au%3sbM^ zILX2CSm`M!$e@d-&C!{vlpcg1`cjO0iIbto@(LvvDQ^ELQOU`KbaN+pn}#<5T#39b z`;WlMDN!hN#uruaJ^|GOWQntDg^nolrmvj0TaeRlMy4^Z+%`7q+QLd-O?;U?& z@q9KkFb%_XM9~&T%~0)(WIYl=@>X9GqUavBzrtS+NY~u+O<_zw3eS`5vIm9%7`z(7 z9;y;rR?O?AcM<q-JGE7&5Y5kzxHVPn6CaC3$qzr|V{;7F#ED#jDB-DC4Sz>RkgK`=gY8 zgA%g5^bYqer~7w+lPFYBqa(Wbn=WRII@Foa@Bk^yym1E#p51cuee);W@3c0jQU>oiPKCz_6Xz*Lz;LTLH}Rr1NRj)Ej+0LAoutM-pD$RR zeuQxAnJl9WP`I9S<#3V#u&p42>Si`B+z=x1f>d@Ol3}cT!K~SZ*cwQnO7ClCLD^0# z&nu9ZlW?h+K>my-fj^K;RXzf(Lo1kebRAN9JhvlpAt>w7blv^lenZ-fm~4rskwCfM zvXjIEoz^G*=5PzD|E(+i7V99E3Fl$g0Wvv38XSXBsMAq(u>g7DZ ziypED+lKi!U%67zYVLtF5@VmoH}O%)3HBYEaOT{8^GOVO^I z@H%pvgR68epDMW}rSTD8CX^+HdMm0oaDzAKI>hc{?SlZiMPMy-uWye8`&)+}t z`zJIwh|BT!fBT;^+|M98+LK9weP?i_v8uCvfhfr->H%GAg`sYq|DR(`K{Z zIRggQm`}Sk_&amuLx9iHi;-d;2pMi=Ipf0Ur zwK&=c+DQ+PLXKLpl1xg=4x5JV_-#eZfH%BJ>tOx;N1365Uev~s`;+zw_%&Kpia{d{IR z@GwuMVuZ_b*)dsbOsGtq@(6R`lXV7wHv4XMpa6NLy|m=OMZObYoud+BCBeFHVGa;* zg-8V><%^q#{(uH|0iQSd88V>PjIQ%0l-Dv8QTZ3`b9XopC#zRK22#Ho*zjv0M|%z_ znB?Pg@~2r9S)_oM@br|~QMgZx(CrdXX@DEO{l$}=69GeW*S8}w25B>N##`W>&!d~I z$HTdp)XtEv-uPGaF|1P-z27P-TP*OyK{+HLIi;q?7nAd%s)%CIJTtGw~!!icR6(;Xb z==Vz#f3I8>6gGFdHXbQNl9mX3C~F4oS+MzI;W7yUD8GwElbux8RJkn@y6|eyPzr_j z!kjg0_kp)$C;m8yV6RaI)E9FBa|NixjN8*5{qd^7xe^mRM}@+Qh29!~sV{}Jq9$8q z7Jwy#NN=PBbPNrCHSMwzo$ie%Spot=1yi_CN`fAo)y;i^TFMW11r0h)`ca~5wJhj}3ObLcoy6snkedrDcxqJ-Z z6`yO%{;mC)^$$TjV(G%c^P8Rh0=|UQ}X2tM67chD+DY6`G^&3{?PlKq)wLCEc8(<3tBt@g^_0aZH3!@kar z72M}Rj=;&jem;z|@@Eb4!IxZg>Cdj=vm<+CELc~(Yb+#sAC%$r2C7T`@6B2EgNk2) zOW6$Pm^MxjCL*alTY)lC)0oA$_0WQ)7fGSF5I~oXEQ=noFDrMdJB-hY9Faf#zTcf# zj-up8l%C6ur-~$kYzDLrGYV^9O(z`Kn%~iQUBG~6-MF#wM>W$yMOW;T^NlGh(z;6y z*DL#%=E!7+Ir7^wP%fZU!|reAq^zAmXi|qONY09q?<;Z*I+JU02#DDjRx?HY<}HeK z{BD6Y|DL*1Qw$=5u0?_gWscbL#^qP%vYU-ocfVT7)2pSbZ=UPNush!YdpG;_0ybpR z_IJ>ET@NW8uvz#n$#iC<47qjQnAIn?x6}7FHiIbr#;jZ{cS1%G;yTo!>YCbj`;_%g z0X|!+U0w%m8zCP<;SxP)11@q<dV zDGP|1ijCbiF>mRSTi{`X%I!RDaYzO70@U{RX4Lo^{R2%OxoK??rW8)c;CDXl+YQzo zG5$M0(9dL0M@;YJ#s@ZdmVlA|LW#3^iREk`&K%!C{Yn@5aP3^kH*R$94M4f6H;>GD zMb)wT&BEoA(UNqBIM;TUTWsRG{s1u|z*3R~>VaOQpieXk!LcrV{`N%=iJxWx=qWC3 z#P;g7;l>PabXG-xugPUA@y@Mpo?H3F`r(<|`={yytB;5~tqSS~VN6rEDf|8NLsqJc z8VZli@$%XK_Lv!>|7jS|gQwGi|DMn)r`RTfh-39;(ztqdw3zuDD?DNsqtv@#COCfL zMLO6F__V&X+v$_!eD&c?I+CZug4vEZ1RR}=6DfHsnG94Ci}fZw1k@>ZkSUG_Eo~W- z^Uqq^a$jj3pT^kxtg}+mO2CR`*)a`N-mqlNU9rZNq)yUrysQ635bY*v}O!Ft9#Yhc&~c(C;s zgAc0F_@cBt=r(vMSJUxrGHMBTwy&a;D^x`~ZWusSvUK{kXa_D^XuBK{q{RVuuF0Ww zZYeEu#(j7d;g0x$4c1>Vt(EgB!b`@lQd{AaZMg5+X=fqO%FL{$DcAcc0Tw3qOs-{v zoysZ+c*DpblYoC-6D|+9nuT7UlW$U7!A%!geV!ToeSpA@p~O)Wt+F36)9UoCrgYa= zATH~ZS%44!yP@=f!gn+(?%(>mQ98ky!av9R*ukF-SQ;y$|Ijr5M(F`3&)2)1(@;7DEhd@%<4jS6oaePgMb`6cb*Hy00!IzFi@>gcN>*${J#3Y? zK}FouV24sh@yElhzny|y#f#wjQLQ&uTekvjeBDUScdu~=(;3nLcBP>-Rt7|jhJwAL zcl_%|jl=18f`I*<+RK6fKso&Vtuz}L+qXOy#1&?>%VYFA8P)6?y9$e%oRMo9AF$^e zFg-LXrzNT$8fc)%HA{OKwd3@%)5T$*z* z#eEXiX%X8)sm3mS8ngm6>4FbHkh+)WQ`7g2&CMIL#FD>g#B^M1WCN+R{$M>2|1_$f zE9c)ACG5?7XP&qH+RFjm!Vypd5uirp3Y)GoO^hXUqYP?x8HTv$rPp1WvFTgp%LY@lk!?<-wMzVVr( zjou)Kp7FV}nB-MI=f9VB%b?d+Ly)#Lmou0+e@ZBNI?H5(^q}PG_J>7Ij3W zA5+GO7w9=`5#4XWy@nU0`o%p#Xw_JW&+fwA7b@HK{nn%~6NJsPuT!f@*r#t*<26$^ zx}06P+4G28O3#3HjVI5c9Cx|KW#{|q!-B?lkt!sD7pZtW7v*MSJW+VmMQLYrVMrQ< zGKY2beA)6+Y}#InrZx_+s4Z6_%1Q9W>u6BK_HvzCW4pKZ>zakm#ig*q(m4}R!^`u# z!s1QuyWA&#i`*%7;R9!h+&k_Fee{O}jacq`L1-P2j&8f2KbT-SL5WOUE$#7hB#Hgj zBh={qhwtUeI#WC4=P6~1rfzSS=ouHR5vL4$Y|4?0`KU6bliyDEKDu!qQm}8n zk1BEzuND0*Q^W#xo1=ax2avzqu1`+|}rhMaEzfX~vZU)+26e$k4{$LOhFAO=jEK$>nL zrh6_SU0@uGc}d+Chfbk%rIz&)1IMQL%P2qVgd=h-*8Gd7Zf!-7e=+Dywh6$#wGmL6 zS;E_#fU=#8CYWm>#31pydG1V*-BY=+@W?s)ASXzYsaj^R@%+Y$9nCZe7nRoTzpzqc z{y@unN5N#H>}n^S!95iXEp+?$`qI?L9Z+8+i@$RhZylNX@&c}W^j0>ia<0yfET`MX zaHzvX_B{W&>7|Jq^Qn9wG9efW;q^1TBhD?@(eXk}@+GAf@k#ddMsH-L5Z6w|y6(pAhC_xI* zzq=>}nVt@rkU6Bl6M6EdaurClzqHLR%J8~&?D(4Ti=BNa*J*IZBl`RNpE+X)z$A5n zmc8AxUXTK0V_z6iC6jI64}yBz8tY$x^y~lpg(BPRqze;oLhYg?RQ_MTkZfBsNP2cK z7>=^iOqyTZnSbnlk*#{UaB@vJ-rxfW)6ya=pdiV{lqiH?`qM61t7Zc0^PwVe4@B^` zjS_V&1_ykoe`yp{GVwBE*MP%!P*CT8Z?ex=U$Xb+K}*K}Ni(li8KGIfOBaHMu6)>F zFaEFNWUGIzkg<*7!~bML^sQ_#Y(MfCNcPVLg{s2Q<$qGtYZkKdYULFg+*D3C$(pNf z-?4R7x>`Q#ucq-YOA-xU>6H4e{O9!OZbWAOY3cmZO>NVdgux@(nzEb!v9EOnbFJdIJFshFT4)6}wkK!pE?b5&#; z-9(Lw9~2|Z9a4EkB`{ljk5_C5+~9gnj^+ws5mruTNN1*lfG2yc7IB^^9ReW{MO z>VDfa8md8jy@C#Gwom>z7Zht%ByY+8kN(ZCtC!V~GEx-o{g=K&b&EmOB@>s=`vC9S zfu~b{-fK`+n-dn4Yn|3$+6e)nHz`jgSTKvo7uoi_QBrR#d@Eyl=0CZVh|>EHWs}B# zzI4E=iWuG9@K7M|rV%4QwSt&fM<{x5$5(x9@Y`Yz9sxahlyhZ+cv@j&G0=P{Jb8BlORfS{Wc9|9Gr(k46B|ssL=j3 z#@v5P?X%D_*c9|i4Lc`Z6b15AO?|B*(q>IoL^y)VfieB$@+2J~AQ8P?zJ9Z^+9R;T zDYA+X2fr_FR`z4!K9qfC6nX>t$>~_qCz+`kg8o52AkMBFfZSOOrW*dgy^CECfAB4g zmhsvHnPss*7=*gP8H68@{J|g?+Z3=HObGn*;Olen=f22SqG0z!Kgt#?smqmtCgAA7 z|96F6XbIp2L!tfurUlf(SuKOL7Ge}ZnH!;~wHETA7uo}mYb|76#X&7Reg#D{p`W+N$+xhu0D2h(E4WU^&uSfo56!5<{&5xzr` zAI<2QYafWfugvg>$epBU2hoc5j1Rrw;BnD|PF7b1zXUXA;|<+yweLkh%N0mMYp_!+ z+-}pmOc5W6ANWZn(0{!||3r!!NPHudJqb4(k^ElJz=ts+04oK4iRfKhjm4oMj@5;# z=<_;udu><9_z3(EhQPq!M~}|>hS-3E4>a>v#1?HNO-%Enz0H8LHQ*k@Rwj? zN!A$`ExYqQkD%zzYmP7m=(xT33NAGe1AqS@#vQ0Psbh1%mIr1PgeT0xU%)SC-VbOE*)@*APgEH+u z!P%fkYHRR)O9Au4Yo#jQWby>EMrV1QVUL=2c(;z!fBmy=-x>s7jGoQl`&w-!hgRh( zd42oXo8GmT{^|P;04#u0C7k`UbP)h({`|rbolQM&A^I<)Ncpo)Qh|ejihbG5OYG7! z|J*GLtv{R421fl8^SVP;`W>_X@P@;ZU*O2?Un+}y$ovH8Tw&&Q4)*vJ>1*q3gDbOx ztP9Dsa9p@wpz1G$m;l?iD%?ORbsO7X$5O^Y;~N}jy)m+(e#r-i2){s?l&)8`qWU_2 z8_!u94Fyc|-G>{iC%JM3J`&cqU&*N-?g?1m9af(rRNEqG@iil)ATuvI++=xSb=ywJ zvFJud96KvtQ~K^ax;*iPs7ZCGZMd{{0wK| zQF=B4`owrq9YnSzzoPBrcn9pee;GZ7EAY*8^G1`Q@>{J_;;VDky(kQ}zPAVsLnr1e z@B=vgWpYhpl)A>}vJ_Aq_+dltg3?95e5l(myw%ZOsr6ShKcVkT zflDzAzQjzkM?@$*n1~iw%22fT~FJC|J=c}tIqo)W>pvIG z-oWGh;T5<;EInu!Din56`*|_oi8B)r2rbuZNEUrnT5+2GnVhcCC>8fT7J*?N(E&;GaSW2u+(RR z3jEQ(gS9ufZ~Y9@=uKsqGG2N82)-ZjIDxz^I<>Rul}{O9sINEx&oVO0GOm? z**9cqZ(7P?)7sv6!9JFHf`zD9MQ|obRueu-H+Cc|fifkUUge9dKpzwg2jIFwu{!fA zB?lco)Ou5gcY;*m4#2MEOQ=x-D_F)Kcx&FcUlG(*n2^~&{@!X=Ru5s;0i-1}AJGQb zu`&(t?gKqjD;L@efEMtsK?*8F4^N3oSHuyU5qE~tTb)ASTZ?geUCMs-t!Aq8#vI(2 zo6P>0v@>z-7xr+Nq37)0&15uE#+26e*1L+pFy{1G*2;7yB;1m1A$f5Zeg~mT0gHAA zv85x#*NX}$H2QZUC3EU-H{y?a1KbIM*miF}M|>K#Xwm>Xp;|pAflbld%Qjb}(Hw0O&3#3jR+W+(b1733%Oy=RcL-?FEXkBi zQ4uXoMG?#e7ZCUksL#~QGtWEU^}T<7{LyuJ1!uX>z5MRw+~*Ka|JVEY$@+j#$!+#H z@&h;fub|4mH3)S?@Qmq3HAkHQZ&ZwJsqDS*^fff6k6lDL_9|OYa~)iSDY4d^u9CL9 zYTJ8$_Zd(lcp1J}?bIY;&^+rHJE>F<2QDu>!CR&J);msY9>Fp|$1zFgE+P=79S8F= zA*&92ZQYI(NI1NCrMbDi{2tIz28;5Jeqh3E&T=-6U%WEV#h<>zs8gY zpBiMBY`P(V7wu$)pn-Mp-;ZY$=C2sk-+=4uuixPvCA}t}*i#zyB+KML5w~`kI2<$% zY8IAftuSA4Y^dIU0R7>c!~l*b=0bDL(S`W@b%7q2xfb<}MmpU&j*?9Db`U6XvOq<< zF%uPdbWGWll+*ivSLDp+11-Vq~y5*Pkx8}EOtyVz%w z8ExItbTM={-5CF2wN(rD$WJQ|w?toGyxg2}@R!Sr_3Bu)F?@E*v+<2<(^p!7%S&1$ z3b+saMq(kPd;tRwyre%8sj8UQk5cKNKs72Iq`P0d=2wbB)*|o0><^pc-5XbdXwP|P zn^{ga|DosN?!Q8uIXmL9sjFc|&q@L3dnvVNZ%0bO93pP!s_2H8eSwjV!E^l|cX$6w zQ~rMPU!sx!e~I70jyCQB`R}@I{+-&hH*x`RdUNMLP2GO|6P4I=v;iuayVwU%Ny^8o z2-|-+ek6dAdEaK6hyJgHWxta4?*{eue`Bh-3BUIc{&xQV z4dwrTTU1+9?G{O;EKn}*tv>1f3?dvHZCKC2TAMBX_ z4>|v6b3xW0ArcHHBXhhWw-YaM?am#-?KT5{=?TVE77*kh8J*5{jA zffteK?SEWKeh#g!?bQP}Xm2FNH;V3;Z8ab-u(YEfVs#g8f^ONidkbW{WaKJb{FOB* zN6V!fOlYFquW&_~=#TF?)J@_=8(iHyUR_m@$*xizr7L$vsrNundhdsuw?IDZUxeWc zj|Fa)3EOISck7=!zQdwAnZ5lMJ8mJ}+?_Y%g4RI0D(919S81zHdWSK^qj%_!%m*Rs z#(MrBU;OxNyj^hOd8g$t;s}g3ZSce%KU%lend^IE%!!AnXc@QMK;x^AFG_aJ`9Sfs ztDHg#^4F5+LDS*DCaZ>fH)c(Ecvl7?dh^n~4R2&9g1556ZNpW|ZkQ|y?|cm2j!)Hq zz}72wdbhG1NYu)U^nmy>Yj$2neMx)v=;Fm|d4fcinas$E|3y-*o2aR0T-;LByh5nQ zn8z)#nJeOB3uKtKXi@O*(Fh88pZy?JUHs^I^H6K@N&c&?{8vo!KF}-AY#yrUVOSfP z|GW@gO6;HFO%(VS(dnaum=c!Tg5mn6U1Y1k&RS=7UUA=({2JsCqxXDeOviPdlj2J@ z2vL8)d>Qv{EQ%fPo35P0t|=+&dxflVrTIDc?NYWBR<9%Y_7Q^mR<6(*nAlFcD)aS` zxkvU_H%a&<+Tqjf<2dMrWt(4Xwy7pc*z@>Lf@i-nllxP>`Oj)}C!MZxuUUpbD0&1+-$AS=qRsq zo@f){VuJAV9$W=igMKe@XQr#-5TcfH?qqNO{Y(#fTC!SdJf0CxvThko^)?Ovx0G&u z#>MyuA3cWOf4Vg&`1coO(4;|GE$}E?jfjU6uX@$H_o(nyY$4(}GCV3_!n^)vEu$C7 z@1oY`i+)ON-Y!`iWBgI7?(sad%BeTg6B`a#f&%rE=PeA%9@Zd^mb?5+Ta6xM1V=Ig zzc?}izLdG+SwX_ei|{g}X}IFQWQW!@sJn&G7sx0fCG(awVth~KffrH3ztF`{CfO)1 z>TzVM23X)no-rdJYorWNZ*qjve4OG?MJ~Y1tP#+U?ahZCm)UciqJ>k@{exAdDkxL# z9NhQV84viBQAAgz)0Gb)MTIqz>P|`X7+6pxvfLuUF{panFL;t1N6dK-;YanJb!RD?u%M%+N681 za39|NjBx!1VYwCj!&U1K`}HVMnHllr!G$LW7}($cq9p&WGA`z!Y2E#nswhVv#CsN5 z2j90nXdFh%Mg%;ikem2pTW)(o^ja<)W_TiM1;~?QJwnu0Hu)iabgbtNDFn5-tX~*0 z-s8ZSc%JusFPwJ4i@{0Wvlf}v>?$x z5|u6^ldujo_1GNIoOE9Uc$r%xbvFyFN=&zfB}>)An`ZdhmoPc$H_z(fpol~kwe1gg z$zy7-rygdcNJ}zo%jR^4Bz_%iw-9nr`h)1!ICA7^pj|BETX{e&R-#2Ylzn92<&OLm zpD(yq){j9~hI6W=kGpwKMKaUD4TVGx$+K=>E_gW)f%1&))aEQic?Ed*Vx$=Sgck8B zykt<)2AZ?WNhEnxay)1coSjd3lbrs%!sYOgr^5Cn8D1nOo1ORCjI9@9*b(fh5xMVYYOYq?^=XgFD1Q=6JU7paR-P0;A4vdRPaycU@JVQi_+F`-$)A9LpTKdZ8aA;{Z+ zsos*aOyI^8j~onqdAS;K{4C@rhZ|zp)$qYfG1kA)Zz}Xd&5~CXx)4QGDv?LF_sgV-}du^NH^v~O0kM?^>Xh|MgQ&-Nypifek zz9=B>dtq9>4E^Pb@R+#}_q#Q1l%sYd2*>MpTb+0vUC706wGQ4Vj&MBIq!^<55FR_6E>`Xko!QXTZ&OH-4hcPPigF7Re@F`#GF7XHZT zSaN~R{_J-!`8d=(D1UyZ=)y{v_|Y|CxgudlX4EgyZhG*wU!dd${($)O^|L z)(RwY)S_1+!7AgLvN~4=1tK;XQxXhmKa48+U4_jPMl?i}_L)OGT7u7SEQzY7K#iM6 zACft7Wwwlb-X!vLh1PtTXtnkvH;4(#WD`JAo?YYt+Mbq02hb;tr4SD|<2G{?D1vqg z(T8Y1JGnyUwT!|Lljy#?eXlW_wX;RlQELHzMePVTMor-n9u5-*gp>n}FBv&VOh;DI z6Uw!YWL<{_y*78T5<{2x)~o~$#0J$0TNyRQ-_@p~(?{0{uMfAS@j18{ez1+s955^D z#w~q+Ra_uRYJkuB{25kHlG_zI#|Q58UR+G)5gLNQ-p_|B8Ub6qeF3-#Q}3J{B(}cZ z>_PPshQlM2Gx(#tPSchP183?1F0*cgF8$QH3!U=JCzWn&-$tfu@P^52>+s2>Nq_#Y zaY;p-mLAH1q6s|xG5dsG8K=z#X)6p<6iz4yjk|cHc@ay~#-D5P{Ou}Rjh@su=XptA zZ~$w(wsuAntuec1JF9`t4o)Y3V?X0k0|n0<5YcS67;6NJZRtbtu0kK>CKo|pgXhJn zbmJQq$)%IB%Xa%)#*tiwvXSBr97bWfdPxwxgPpx=&lwfoL{A1Ik%UJ3{oa2LR_V$j2#;<`=~5N4tjzAUg$ z8w8!|ZnR}XpQe68ymB5u+YzZVi;N(X3z`uPZC6(in!3Hxh%Wc#TE}__1rfIFYzyy7 z7Re_(suB8OEK<<49M(f0qYABuD9F|$3CIZuEOfk`AL!dyHpThv7{iTy3fx**@FK94 zOyP#M=V;W_SPx#(O-o+=E^66%e=+mckRXBku6jA3?Drz>zS}tK|Uh-GhGmI?|3&fJEpA86ir9ef2)6j7+Y|b;7!4+Qm>t@ zse#9H+)Fr0b_M)>O8rSQPFc2A>7=jaGH9UJJ=^wS-`5|^XUwmcnAo2o8ZC*?UGhFM z-Ex1ze3=&xQ_u+gYV=XCvh(6c!9t9(_z{m_j=*`Pjc=i5n^UedeBGNRt5Al@1J8Ky zpA~Q-)}g-~GQQ^{_Bcw4x&?i#Neir3ruT^!3#~m!C+2dv~keeH&7cMI^Re} z^;;OvV;d*Bp_cV*-wH5x8s%*MR%cgWD-W^CqU6(cS3T{a!&4r~F7Up~$IojWwu{w0 z{;lsdYyK+3)y)Wou0~@3o8dipDVC34h732REmGk16)@tvrgrYj@VX!U{7&xsJ1zwz z+VVu?$`usQy%s3HBfnkjjFdD$D*tkIHT#eBQBt5YD(|PDpYP;$9|YcPl&R5Z9Qsah zlZUvi&uls5Fvv?AinkV`HvNs4u&vdF{uQrRc!ZD6fr zitQeo1FpwT{d!Edxx?%FH`Q_0@=c7R`tZ%5@s`YkN5EGKIr$;ya7gIO*JTg;qITgQ zubElyH7bBQq0=B25HES?)m>6&+q9>S2-*-`sW${Uny~)Hv1!mXNX0Z*^l=#U@7Wef zhOucQw=qDpCZ}!DMM26BGd@jJiw9NP?*}t`KHsjNy2CyJa>`s_rQbY5K4;*dkH@yq zv!SIsRK_255-0|X`CV9rlag&`jmC-bn{f#A(=exI@q{rCVIfCc{{rD4*Y9t4rSlpg z2%_GCHy#YX{ffY8zlKqJdrayxkAFZ+e%^SUmMCegAY*W3|RKSE9Zp7C5>OwuTN-s$+~n!jEY0z4ub{Zt5mitkTXiZ?2@a!GWJ-|11EOJZ#RVOaD5qXEv&hCo)qa(- z9{PMrJ>y2^bKVCzejBG!0>Ta$a)TxbIi055xX=&-e7a^(<uL zdLYVOnOsJ!Ue+V(XJUVaUw7WIxws?qV7YzoRnI08f@qx1eI=I%x&A3A<8@;951Xh9 zTwQg3!4fdf3thb~USOcEtAnrex6g6HLi5!%6BeUghE-+VMfE*iy;t%CFtbX8QEtZA z?e!=f-INf!D<5CQbZy6z$w-C~{bWax8G;!+psBBNdIkM8ZR$v8NQ8z2s6`hR-xet2 zHexm%cV=LpI4E)_qIy0kG^AGcS@krwT`|$iDwzM+G^FmyVaeAUYhU7A_A8v(5e~XtC-9#7dy!%`|T-S9E19IUhkdO(jK8ocU-ELd;RrIR_*5P;n!0vZ$~wAhNZAN?wQsl zm|V+ElaZ`xpFHevgLo76aQl$3Hcj(^en9=*+MH5*p6A!VC5K23W*K%8e0ihe%{hP1 zKoi%W!8(G9GEBH#FDG4n6HJ#*V6OO6LXVrC|NWpJpW0a>_{1yK>MBN{bSB!vh!Jp? zgB<^EyPjJ*g*SCA_+jPO)xmH;`91^_mWW5rbF8PP*Q31O*s_)!78d=~1{z1ntWONx z92w}oD{Rlb3HO~Js_2Z-uFC;{{Q8cT!qd>eP!4aXE)sIhL=ncyGFVzjxgC) z5;u)Dsou^xv@zkGF|8V~OeA0#-D)}(W;xNC#)T~F(YO$si;@oUVdd>Iq786*<~}Y4max*gMX2z(s|zfgbbar-$P`O^BE7Kw*~lNjopKEm!TKJL$U-c`b+x8I+X@*xDbakOiD z{f}#`F}^(>((kS}_Zu0Rf?}}1W>#WdEMoFFL;#OfxjBUGt)|EQFr6X(#AUNV^*M+(tW@KWPsKGz*;^&-2phX_HrC8aR+xpr} zzq;;IG!#H%X)I@c&G%0zHUDG+G{dS^Quw})9z=Mo%}jltbiy;$;1C3W#r4Q-GyJXC zojDn$W2)%x>N9e>@|}~~Zy8c)jxzeGe}t)PK?eY0?vq3Vzdq-_Ys!4jP>tJtOh*R9 z;$i~6*1CC(ouEDR)c$i5?BOGH8pmzVz|~*oekJqbMC*O4!p;<{^jLiGmqX<(6T9Vo z6-=-GK${4;9Cf&@zRrz(O5c&^3Lwn~)^u+bRLW`=tKUpmjnuvTfjGvwp6YP>@=^zk+Wd z_Ma1soOs9{%+8o`O$+q&TaTGj5v#!7n`;4;LjpI8GK*zZ2iz!@!#sC+v1S1FKR?A$ zEP(kTbokU;%#bRdYF+lYeZ|?A?USb}H>iKMxuihzPkg;jAZOdL7p%j5xJfsXr6?dP zA$)5w3uO2K`wUd85ygh#dNXYAVD0nlbkDCcvb}F|YU|d|)1PN)PCdQK{{4CPU6nh- zH>d9-e#>+o6y!~YBVCnb8a@D(y%x>kvI1v!9^QT~zqU?B@%>HP#MGVhedZ^{B;_j= zrl_SqkO+Mq%MCE8$*1!aR{CT*wpu3oi}Fof*f9NDyb5+^4`(o`-lr=!ytq_3oaev$ z@C)k~k5?OwpH};6pS~hk$*HcRm3p0DUP||2qRLExERoSp>PYMJUyFtfv2M>W_}QR} zcV`k)=V_;G%-E*yzOa2^fxkkT`6oTe7<3v&vnvVkMWRF>0D-Z`S#(DTRo}@(_XMaB z)>I>!48!-$5@h9H36kG_1*nX=o3O;!L4jSrtHzm>?Y!#}ho4~#6T)ujDjA0;&8*4o z?c)b#sWSans+_4@=B)Nx>PF1A>ujy6%f{rLckjo~vUVj)Im}Py#X#leI10F*ttKT) zbWg8_^J$;OPtW>9K5oAue^KFSt#!*9J4U<1sr@eWX22>J0yR`};S%N1@4nMSzEkxb zIN>v$b#~KDK19vg?YK&|)0^G?a6au>(sj<~Sw@Ms=+W?w%|-fXKUONdpsvy&(I_wf zqftDw3`e?TqIrQ4;IJ9oxS5kmN|2mX;*_;-IkUD%D(;s+_QGHRI(hM=4{!BVDpm%Jr|%l^9QDT7aJ> zN2*v8XhmSlGL1%Z)uzg22{Q%4FZ?gg?f6b=uEtN>LaZMD)4qQK%VW}}D%^qU1G;A? z0ZKN#1K9qo>uQT(+qdzB=9;riQXTNG2TiG1z<0P|krigD*xfT3eiBG?^ySb!ap*I_ zf%^$_>OlKHgshDH4VUR(^RV`H@WojctjYRLrVb*lH1rQct8UJqubM9QUckjN{ISrr z;`f8B6M=hXIZ)|e51HL{coi6R!W_~a0o0BSfG1L&A+zF3;9BwEAj>WA=h;D5{OdtG zs@EZiJMTtrlbA0F{LcdLAGAW}v}%ctw$Y|o7iU;Gi!%*08eRKR+fRx#zf2ovcx_`r%ACSQF zQ8eBFDUqV5`YJkm6l`7FkfeU4V}qJ#s_k~i(HVlC-8ZKvF8M1onIGzjKbrUt+G002 zqVmI@Zl2(XbjOKGC^fR_9}@0tQi}gzF+ z_l@dlqU%#4dVB-0bx#(pVG6xUQ_Hl>WH;BM1&s(%FKD2bz7ZpSjTaI}bXQAtM-H~S z!jw#Rj$0tJLou9QLbaP#sK~OIB~b7zfnp^Dg23CEb>&LN1fDR%C@g%GQ-f#l?~MVumor34CcTVLPb6KjPWOeEL)SJFa{@Nu;PAUe#FK{OyG8q2MnmZ z$`VpUrKzyE9!YdTi9L1Vv>1WtQv7(HDz@+>ok?onby^j39X)t2zglR@sDP+h-}hzL zbn?rrlNlkajg`-yQ-AN7J=r3sHRRV?2Gak6_`$5mLN!}^$)jPbsk(UIEIPlIq8cW z+i$yyXV|~wi=ugMd;6)7OZcxdq{t_YekB?{!rqwv*oPWv)ZCATtNWip3?D*dNM zM*mP$@Ol}oc$gJlTVI-+>5}98d2rtQA~_$R zq55P{&kx^qNv6AsJDM|i@=DcrDWVT$)LgznsV`~+$?{-`SlD4h%60!KWIyVPd06>E zR1>maPj;ec4KKjprb*eui2bD9*I9ajh-|-!lTAacFRUmc_Tbj~9Ot!mEOL`+C#Q^n zwH?}b;*R(OZ?uAeYUBm^7Wb&d?-H2^CVWa{Hyr~q2h7^~*D^AntpU`@`Ix=v!;Fv% z17y`s`S?6eZ_f`b)u4E-P(EvR{0wePiwRj6*$F;D=KQcR=__X=~d`f z3)y)CNbn<;Um#=|tG?P5vXC_;7ee{f_>Eh+*o22m;9iOu3C&@8qJ9Q%X&wx5uJqmY z)gtuH@Pw_n?ojZ~D;I#Pz14KrA%%m$(`T-sGimEmd+J(jCkK{Z=w{o92# zjMJ3u2T*A|Ll|o$++yAYS9V}i`N@y&4G&3$`Q-Kzm^}Ee7+rm&zboCjg7YYR_zf3zPZXOVwFvfep$~36ucVM7lRgg^YOYmdJ;$!-w52n_>mlwRty{(&?=~owlM1zu-QvSn{7RC2>H6^sL1UDv zwh;9MazyEC@IX!85K$J|dpDBq-u&wNvlj>U^drjZ1ukU{YwZvq>lfm8u+o=XnaXRy zaCuveRXyZuOG0{6LNec$^iBXwE|C=Xc~!1)?20wU3QW*5 zhO#93U}g9Hm?XT?v;4vFBiPrEV4r4%XBHIN3<7vSVs&wS_Fk|vCxsqVm8ZdPJ>aNL ze!gC&xF`_qs!N9-x1;1=dR*SbAY#-A$@z`sW97Uq1ew4AHx^gwIB8+_jFoDaXaOaY zIcBHCL$ao{RO>VGOonR`Ffv_H&?{2kC}`T0)-+sn-N#Rg=TLg}=26B|oQ{=en=Ld4 z%Xe*XwJlTq8oXS;vaqI{VPO=wy*@P^JF**Y>)5apdHVp$nTcBUOv|mU-jeX-JcH`N zbcYf!Y6AiSqsD1){Qmpjno$?-0&D%|aVhX}Vz`)eOxmDY3~4PrKqGtyVW;nWR+{0P zSg$D%O&KVeJczA+1XGxy$C6pNl-6SBgA1osGnvO!LGcm&{JevX)5mplU@{(?FGc7t zeFR(VWM!wDHAGcg26qjDW^vle1%DW7Pw3z z)=Ao<$zu4zoxXm`kWoU(rQstTE1p8R>osvb+piBAG}L&ohAYN)Zk(l7_YA(0QtMu? zZ*vXuQzC42vB@@fd=o~kef;!W;0LMD&7{yODM7TFo1FDz_$fB9^$<<*2VaW=b7{2( z!m5eeJxrQe0ADM<4|npm;`w<*|C$Pa_e*mTFkKs1xRq{)@=OR!>Bfvx7~*CDT!uF1 zaEoi9aro@&<&XG%Jxo=8986v)^_sM?hPVL0d3)7$!c&5~CWgH>f+d7}ygv(o+dcuG z7I<<-3h>yCrLvoQbqc$^0kTG)3=G<=_&MIM=gq>o&|R95T)$tX&1|4a@B$m;5k0LC&xHdKN>!V{-C_>>LS_ zaXC1{?s|U2a`o;X=4Toy^%n&Z*9^zc1Y*vjXQ7mW;orQ)73_;+#~&s<>siPt7%N;& zHz!V6?cpJRX9AFWG+-_$@Ub<;$|pwUhHJdxdC0-RM-}OPSps#9?B*21H;Ee+<|1@K zV_pwy=p6q*|bK}Lb<#4=CZk{qJg1uzh7`8+gEXB*668O|3@?UZ;mtzYeMcy%)RfO)>O$F_mTq+ z7Wd))x!}!{>dQuz!*47+pRes?z~my1(iM!?cV@g#eeuV|A9l*k);VRO%Hwlq!V70_ z{_~9cQViA7j_YXJd-6^^ysB_H;D=pitz9j;$0`K-4E$7M;pY;gf*tb3rhIdf3AwRd zUn6EV@a_K1>|qUqV^w9LCq)*~%zvJ+WnAJnhua;by_8eZ(K?#uZZ7H+f5^Zpu~nfKT45y$pJ@fC4)K}q454^RtgUJIb4gd zCNUwWINZqTalKaX5Bp4M1fr|<(ZbINVw(&4!0qej$yQX7opN|37n4DE+YQ;U4pT2?PJ`b-okF{DFbHqPt>^K=8eNgYgD< zoGCcd(QI~%$vTp$r~u}la`^|z*-2yb&a0v(yk`XG^M6SET>vVykw*)#+rQHwA44;y z9lQ9$CcS=6KZ{yhBhiwtX?G>GZL#z-u7^#)G)I2D)xNW^J;kAi4BRvQkWq=tn!M$w zUDBv}5pBLcDT&@a6l^f#ZM3|TL~Yye7#1D+GmfTWNV4W{x7#mokY7wg(rP8v`A>Oq zhz=dW`dVq^_}_k=IAwciMTWl42E!e*DZ4;9$ zEVipcDJ`CR!xWLAUkuQNbnnX*VI@yOyD)v6;5Or#oqw$9fya*-1x9Gk>^xK&L7a{C zj;?2kKa!yL#?hC9A`A>vo~#U6<5HQ>I~%pNFNm33osI1Z-7&z<+UFX_#dh4*FZwt; zSsY0_u7OOvJAQj(wI_w|6g80@{xZ5OImEt!F7ejST#f4O2>JpXc;sj)ctj?G^AO1s z5mAvb<8<&cjvWsfIOH5QNI(~}%Gm5rqGVC8ppsX|elnDmp3L9%czR;8HXlC}I!K5N z5udJ;CXC*X4O`f&8AF&Vv&p88&)>yppBIifzBIVKpYyY4xrI9ECfCVd^W}Gg?ZJWU? zf~QTfhrBtvIeLMWII=HKZDF}qmUc=mba%e*^WZQxfiRJ!ZPr4)blI)V5OxQq7Vg@v zw&HX;o$H9bu6WH4&c9@5Y&+uZke!KqUjZ|^5{Y_r$g zZ!zpR#QnMAdk#D~tU7ryFhciY1qi(WQB$O6V!Ik%VTiy$PWa}{ap36llUFplWy?tq zW5qb)(g!JJ`yRM{d$4GQQ|RYyL+Qsmht6Zx2U}v~O0*8jE+1f|l*&95*vJ#6OR85q zP8qMDBKt3(R0pz|8A9!k4%N~e(eRs^p8lzeX)|%?7cElRosIS53GA#s9KWikME7ur zMOL@XrR{%Wia7Z=@uVg&>RR#1T@k>*L92Ps@6&URl3lBK9e;FwsFv88*wck6o_6NK z_=P?Z{hI)k4T(r8t1l_hI?xMOI^JoklgsMpc#c_FvI*B-#!AyA2-Y^LBshQF>8l-g zpr`jpeJs0UMyzh!{Vta=y6xjKsy&Wx-c$17aQWe^-XE5>+-2R8YDtcU2~1zloJd1( zw!r#Y@Kf?Ct30^%R75NnbOZi6!yxwWcW=Olm}w)hStf+m5(Sek&cI^zqxt^m5W8z;ip{FUbHP2Ot?A+`Y8u(zx zo>$BU_g^^tK-4(HELl60o4-@N1>gJZ&p!(0S$n^yN^w4|q1{>gK2%Qx%U3<`al7uB zbrRT^yE^>QJJk1)hG9|@cc#(}I&}QO)kVc#U#=)>RZta+6=7OBDlIpCaw7KYXlcpn zE1?Eqj$U@q5jmZ0TC!UnMw%ZEDLP2Pb}O3Pxhf0inGl&1IHK1jnEVdlEY#n$#Arz2 z?~C={uNazt2)qZ>lWQWs|4AdQELQ3HyR?M>HgCJW{o*FSO>hb@6RIS+&LM~QPw8FZ zg%ugeZi35UjSLCj{wIg29sH5l#hUW#hXZF7d|tGCpH_k@kP;`;Qrx1vCo(^}t{hzsGS=*tQtiHEFCudC1o2u`U zcVVmbeBjMNr8ZjjCkr+JiHT{euF(cyq+q+mPVA0eB!Td~neA)RVh0dwa~=sxKq1q5 zhv`F+mggXb?oV*y4DtnAI3{{kh_>F-!mxzm5>=gZR40JvMfdBWfbVDjG>3J4?0 zemn?cBO{c5>tIg}+9OiK98sN2o0(dQ#4w13%f>G&p)hV(_?G!H1*X#d-?%yF<6ZZ1 zq|PRUjHo)a*4$|~{q1Sf-4MVIFK{tieoJ4pOf_ueLTYCy$DdnTIR%hVTT#6aeRMdd zzPGvV3Hxz*U4SY`JW7g2O2ZT_SV5*3^^VZOu@7?G{T~4+d=c(0J-->zOZC?b{)B*6 zSd~2e&aDe!uIo=zsf7jDUI7D}Gru3b%p$V2bil;Ew`U^aznoc^p(L2BOF0lj)oM0g%#~{BV^GY zy)}k9odS7kQ*DV2U7yOma$ze-ze?jy|4B>PbPp$sQqh%Rb+*cQ^wHEr8x|AN5C0kb zV!Vp9>IC5LwzA}z1FrO>42@sBMLl6Uv$=H`btB{6^F!o`(1kJ%(79>cVoreU*Gzz7 z&N2V^j&(Ix)Z^{M*KRiiOdP$y4}bYiYQdv2j>kc>Db@-U$H(yPt815ClK9f|3Z^(H z6*a~l87k>ZUu}#4dFhu@6;k9U4L$!jJd)GQ@;)Ut)6+@IeZCdQIqT*eXEa}0i25W! zi)~HSa{j$|N-;9y<3U9|rLWIE%~UZhDP?pm)ghBjev;L(3aP&wz6r>{l3A{BmdgY= zW^PsdpBXZwY={LUi=VuMfu-<}sg!S6@zh!OwII-mot@NLRyCXE79&Avqq3glQXXvs zcjOU2+(b0yNUSeeDqFtZv{6fN52D*F!fyDy(7laGZ-oHO@E_zJt) zI_t_G!V9QaDw)`ADMZzQZ`ZUih<<|oN_fwOn5{Bg*Q0&oCP z7qQNgQh~F3i@;42nX}&i%|#^A`s&GYO_bx-Y?nZbJqeO<&jQdct#*b}XQNeycXmSK zD>VJFj?6XlWX`PqFXC25caB$KhRcil>(TFgGMbR(okc;dor2dEgX2YCvS%YEdrU5HV!q(P$MzEHBRKaIqe&ff4`n} zmt|f8l_cUeI4jFGGpVny!LS8Z}{IQrC4Q#7E0EsR|yt?B{hE4;^5cI%%dckn$Z zrO$v-jg%^D^7_tp%6^$AbNR1Y8$@Tar{Z#g@;#i^7X6mk{VmyKT2&7{piBwm?PqRD zTNts!K#5l?3EN9%I8D*T2mD!z=CL!oS{`f8-8nFf0Iw@0ss4QkO&P2&Lf0VAcINd?=2B}Gb@LfWVey9H!tI3q!-h!Q0nz4mPN498yo)Q z3KG)$UE;8Jq> z9(Q`Yf?roWz$z>8!jWAn--8O?I~nXWm4OU;B0IfQ>lO3YyypuJgVe`k-8e1bd(-Cu zqs$@k5N|e=E;-01l+LN|Y6?B{>$_r3r$^t@%L_Ic5N3ho8R?h57U5}fe~#LdS(tm^4Iq#eg{jcS6x?sV<{ z^aD};jMHP0UsW1gCAE1DCjYqI8W&d9mUa_e9Rsc809mBe_DWp)yp#jh;p=CUY@Jx>O@Vv^M+LO+;;GP2iy7j%+Uo#8ut^#Y#7Ijel4 z7tZzU1K0{e+KBFo@QIiap|m*qwu&zEn#s6XXT8(foos_(=N&I`F~aY4qqN=-ewtmj z&unOE<{~+6k=IM07%QLtT`@MgVZ<)ZmDP%DpgfwI4-#c&nN&t5|F<+n3rTqf*UxKY zrOZ>KA3$k)RWV?Ll9n!0`WfxG_35*_1{I^u$<^{s>-~26W50lcKZjL6kkhBj5#S>y z@8OIK`L<}0Ww5{il#KvwsT=*D5eN>~tK}`S&daK&1K@%AQ6FlgC;qh*3TRcCr#`{X zN|`n=>OdRI|0{Q*8=&Z-|FYm8t5hVvY;NMwrgE*$R3-)aRRO%<{e?R!!Qg)X?$^x6 zy78eFU%B}bv#-7Q2aS^&dd3HLZw|spa|?j`zzwH@uCADb+r*3kcQ8@te<9GENtISn z0p$(MdGT);^*!%;xmixlSTS&g-u%El|3H=n81*)++@Pu`7$McJ-*mKYxgAkismGt z{H~aMe9hFXzM}Mh{JMjQ)a>_O)C}Nwz+R|VH|8n$BgP-Qf@V}S&SvPaBWMrlH&$Q!H^p1v@ z7=XfL{v1~Q&Ks3}ZOnZGu%N1y3}eX3;rMkjI~IS(JhL7*KuF+b1Vxdwuj|oUX7AmY ztDsN+#ee$VMvRkECcAO|(vS$J9kkR5QM0P=8rn$mAw}W$8u!m4*=6!-PGtYL05I~3 zW2)t^XN}$_>R9*kUXijlOLVICf7>G{_*77ns-;CTGHgqXO37rVReEJfs2N6!|KO@T zL_NgN)?!JoYoi=(B=O(Ot^c(vCW}^bRX!&RxB}#P4WF% z4XbH>$MAQv*`xwkNB7Q8T`?E3sJDrtnk-ZR{`{Q?(EMx0sfD5c3k$5D7h+~!sh~BR zSOb?zcbFUkj=MZOwmryHiGzu-&T7A~X%=DqGa+iz4s4!9*%I9S8#pQD+e91Q*%D9g z-kbjT-{+D%tnZ=ccNN<)S|RhV&Vp500PXz!s@}bYXAn%=6ejK+uB0EBTj9T?L~ehU zPVKo=ON@Hq&2h7YU@qZPo|n)6SWEkzh}*V*<{@V4dA}!&j3>agc=HHHCBKNTg%-?p zb=9P7JMLb&*=+2O+?DDN9r;c#06K#t@Ao_0VVQwZ2l^;TOXpro;{PDCevj){1)U2& z)Jt*~E}xqf_ysB`o%jkv|781e*zQXDo?6mkwb`@4%{sbsKnF(V=gTI?Dgw*+V;S!c zT-1UwQM9-p*Y45uewoStpuSHn>8<@=%>gJn2N~UEYv))77>6jOK)-|{$4Db_W9jBu|0|J~BgP(h(DxIZqD7$;$zCONdj zC#PC^#R_P_9VsAZ-V&mjkc0NyHRV5ThqoA4#;C<8{`-}d-a$NrT4cn;9|q+sAeDOD zYrcI#i$|BU6-79yjX_4`|0P+~B%2uX&x+$KC7B=lN1n8LGm zRU^HUi)GSXhtBgiYAe&iF`i?7e)CkTs~V4H_~b-wn}q)={vWd@Ka${YdxRD{qP%`c z_(k2Bfu14_F^a5Bwl6y>%2g>oY){|);en%NjG>fp(j#ewE+u@Gw+U25-)+WzEcNa3V&XIa0$fSc)wt64Ibc_cw}Xsn1;IFGpZBeZTbFvw-_<5kwFz`y?e;&5oYGNSTLf zm@s&mFX+=f$$c%UygofutQZqHo{w*(HjEAUO*&JC8TOc$5xe(tl32PwN$&r3O!8$` zo;2!QZGmk(i_!>BTFkZzOu84OD^h7Kl5;zgk#0rqRy_Jt%gr;~nhRd)^eom8!SJO# zQk1<`GTok}f}dd4vju-d;d46|*>)0Za@jO8-}r&{8VCp9>6p!ok1AxdI}QsCuc*{( z_k`sJ(DQ|3Myx3o7e5wB)psR6=Xt`(Mv8D-kVSqXf@kTz(c(V}a5ogG+z8}Ka3K4w zw2E}=gc?E;8aSFD@S~UC)1%>G^V&>6G2l@_QuR`&nB!4}R~Is=A`qy+`_Nxu!c!H0 zz5`pOYQm!-8K)+KgTk0Tj+^YFTni7Ci>FM*m?zZc2}=Ptn8|bh7;p+&s70@Q-&Er! ztcX!;LXTwmwi*qmqsyk!J0k-b!~Hf78o2!nT_(P!+T>arPGtw)^?zlpsl|rL8Ekea zOU&rMk~ZGB><$0i`TbuU18rj-<0Aezovc%y#m=>LX~w=wlPiGy$}*@uY6b5rsIex8 zE4|sYMNLi3sXraXu=N$a7I%5-cTVN2x-vTVE-yd(;t^eNRpSer&S^U%Hu<7>gkIts zhQzbC;=2W*C_ME-(V@x$pB$G$5z0)^4@#G*Wd38SBJ(PPpLQ^?m7n|7h%6(cK{#rD z@KnfRxDCB0or)6>t>D560tIuYGm%i97nrtJ7_wFHonvT^3R#4pdf9!@>q2Nohv}0@I$af35P^d!snja+Hp;`F$N)jyDd3-JXeB>53p5oYCl zUGK!O0ZZL5NIlfuLYO(8Mfb1m*KZL=N>1fKu%fVM?*_6Rl8DEv2PEa1}jW zuiq(5N6>{^Ro^w>oYf}m_;Vmjndv{3tXX05{P9 z^Syj*JKdx?u)I9x7pI82jS&Cwv^qY7- zYUnXP$T}@zWl8P2(dR0P>n)drp%AW&5Xj2+$C9C#9oc^LoNASD zlL6|0;C#LNPLAIYEaZ#1@P3miqI{1?^wM`ac*W_6^aK4aFN>Bg$r|0monnrMNYW+E zF*C^mCW2rI3RxXL68BWWJAZo%imdC2vMRmjX&AMRsA<9zRC~=OZD&aZvv{_G*`VaD z#|0&;{>Z>>21>~8S@HyEZlKopqHW z6IyTeT!!{4rV?^6a0U`P^~8m_uBImpHEK+p7TU-&k<|rMNCe=VnoXnRlUJfN<%RKG zWy$P((B1pAu!1<$a@;DSJwix&TqqjqD#KfPtQhFtjs9v%(!2I8K48fsqWa(u;n-6` z*^o3z2yai3Y6v=C5;-r6%ElHp;ABXahgzHaa7#}s>S{5shx=XeL8$s=W)QC~0nxnU z;nU$ZwH9cWNn)Qlf?A)vA6$wHT_O8Lav9J;nF*8cQzI_LQKcXzv$cO=y3OK&Dg+8= zBVKY!h_>y&AXKa{XtihQc(coCPoVZuXKcp%%Ldpq=i#LC7wbY!l(Eu}6xhxW9y-)o zsF2P*r-@PH|7q=q=YJi*#e^w-PbtLe>%ce(YKOMurrd|jas(d!7&18|Y@rG~N z5kjj#0Uz~n`a>r5j^K1&{;E7DN1c5*V3{X*IP$;bkP6R>`LAT;>BB^@4Y|X`6u6(oU_l`Yp=cb8h@F}V;1N{X`iKBMbkQ6lLc9V^g##S z>lgT%;iD`85i^a62r}d0?0KX1(-&Brgvx0lI~-B&b>#4BL7P+~L7CS3Tm{k3fTKmE z^%*Xj8QA@ePZwKHrY!voM1rqZs{X>o8D!JWnJY*GQb76O?yroyrx8hnlvQn_RDn4^M1F_EH$E ztkdmhC#m3Xo*gDj5&X=uJKWo0SKG;9>q4DRJxmukPC2{Y>nNDyf6Xd*XJTFIhdSA8 zJKV)W@{s~c*YmN|4qggEg<8)o3aY+Vg{O^1G!L(JAmj)7-UZF*$hLJdc`5#WaL#^z zs*{6IK?`z0z>@c;or&pOC##IdmE7&pq|09m${~Pd8rPjW{iJ8YdrB=+Q2C0)1(K2rG+mZxJF96~YSX32OKt{kmSSkSnsRv%Ll;& zCewWVpSHnEk+G_o&OoR$AK^Yfp!)a0ACF?0h>!MlHF!VO*${Ecber0H>;iw`Alv)? zpV=_Ho z8VALf8-RQVqim255}Ng96`*LuUXa&JdQSs+D-+i+L(|etcFE#Rhvo zuC+3wKPyLnZGW&UDQ*F-ithp8`opJ$wuag{VdgJ-V(p*KOYO{063EG&Y69In?%#3{ zpM8a^jW$QS+b=LzFhg)fW|*#SQTPW(JMP4y1*nAXk<)mAvMWtZ=O*`Ck!n9kT}2}? z*1(cSxJdPIajWvebgqHLlkqLLFi~<|b)L z680)X=?ytKwIM5v5yxPME=vWbFspOEBJ>)HK~}`qLw)TyYW%80HprStXwf8JgGh`k z`YO>9-Tfr1|2)YG+Rf6R6`b$MR2*;jY8f45-kRhw$6~r3))5?BFjIHb}75v4lQBVi0*(XbwezPdasU$ru z9#Hv7B|k}A1cYXJL&jcZi+lWVe1=wKJJE#z*vbm+Y{==kbA{b)8JGYL99 zFM0e zAT}w@+#;v(th>KA$K|ap7W;>ItI55MSKD`8J#m;Zdy8YFJR$Un#of-GPepeiDNDUy za`=+@0VtUX9*IOy&8eC}>hL2HqNQWId1%I^#dmCsLCY?V9=r1y>{1SN>`3pGb39l@ zh))7OsrfvpBv99y=XU^YcS(wUBe~u2V8a3QKFfh0Ah3-J>iJV2pqVMWD1A@rMBGC5YiiXq1_@-+yEeW^k%6m z((UQ9(vIl}&Z;qD*DK9~6|0OXe!X|DGW=~Y8oAW%JcO4sOEDRt60n`B333vMs!UHn z^K9L&=3@;Ntfm;0Z6*oHsvpXcTiGB+5t@;*iYK*&y89Io4M?Jd=PkmXXN2XHO(2U0`~&)rU>WC?7;f9h-7P87mY}bJ zHRRQ?meC&<>2k|*G>Ous=u3~^_{JM!5WvVpUmLHxt8;v=>^@$_zv{<1x?TcF%Jt(K zT@RCkDoB2l9qAq@xvX?gq8aeclMEM{P=qwcI5VM)jwS0MI3I*RAJKabgkZ$ixo>a= zamuKAkjYg>tsaV9V0HWOQfmpUu61mALJa5xrJ;Ek0C^H_J15@IZmF(=OMO3bph<2Sg=x!Oq%4L$Y9SO02%ybD zGPm8SHvmPzbP6E*Z!y-cd$YjCnb<5XP6QjLWfyi`wV51L33Eyn>w0`n3Y$XhE&)ZB z*fEqNAlzSQ3r$P&_~I=z*@w2E$nS-KbmDus@zvccBs9Hpiw5ob9qTB}W2bgxRIdJQ zU1nib#FSB3LZcLG&TpmD)mRh7>$g941ueh|;+&(+jecj{GxvO5-=EC!x(y%yj`1?^ zl2p9nM&gph;`Omh;Motvt0{?wVj6Y{bg*|eVZ=3$Wn4%7O2^7)VwLl;oGZ*hsDxI< zq-?@6O+wiK^hJxkoFzs#EHF_!yzRuc?c$v?6)Wx9mh7D90HsK%h)J!Ir9Mv;t)%Go zIPMr4_Z%<6mxfgB*$-*1!Gy|VSM9G}{Q%P(v!~uXZ(_RC-k2I|Fem?3D)3rs1VVV? zjZnF5Mp$0Giaw<};x_hx4#LX9f70qb=rnmuR?v^Za)-jnAY$uq0sCdC2;(mrNxQF->Dkej@G@yCq-cov&#i6 zpgM(3jxU6sraNil;;LYmwEp#=zOyHl`CKZ44-Hx(RE{`}J!FmptWi^8{qR&7>!@EJ zsDP8uHDl4ddbP|+m7ZT{)pNgpU{^37Dw<@98BEIsP5NV{%qxm6B2(JX4p-jpym&zhWL8x~6ZJR#RSLgwrpibpgdr2rI4JK^ z2UYb(v^I2!zr&F?hF>MJMnw@ld2h_1|bh? zWGJfRkFGT$moXadp&TfsP%C4DxP|q0SLr*l6FdxnnZ6YL=Jygw>JpdjC*Wwf?pkuu zIBcvGT|nzXrU~0y%h=^SmKGTTCvER6JwmxdEC^f210Xi_$I`ACV@s*7p)T;Lv6or` zy&*%Y8OlYV7P^wV{eh#I;^65!r+1=6VZkZE`XECq-p20tYZbQcV(fXaT`6~>H9*UQ z8`K_@%+wqG=~kfrXzHz~w_Ml0U>Z52CWwut`IM2F?od9EMJQhA;0-8HlqMm6_3dYb+yh0-moopMu zYf>oHljFR??f91>oemz3vI+--8zxhZpj<{ty^pC=0hGxi5*}^A{|!(>%^SC8g`BA) z-(cpjPX+BkG4l2dhTz%TfB_iC_SsC%6x%Sp+7*p;g}P5mzm3`&uh5ZxU%rpD1VbKb zx)nr;(vQoL_g+&>e8*K{RiWU0J}HKLn|Q;H@eKexEY8FVG_Xmo9p}5If7EFc!V9J+ z>P$c_ij=!Ft0(j?FT8YJ{T6|JVu316PuJgJ%Nw!|V?<3F-5bgr)aNW!^zh${!vnQu z4kP76fKp&9oS^BJm)#|`m>4y9@zA08y}yPBkYGdikb4mlo0l;1hNFbuj9rvR(P$e)8jrj5HhAXf_XLZq?N1UPD;E zb69i4g|A~FRx|H2<8=F?6N;WKtP@JLJ}LMj)RK~pf|HcNyS(#<90mhD}F*;+c;gH@wrIizDD@RGcK*l3D0^Rzn7jYH0i=3@ZDm+(phBjK$N|?lC z>t1%#iEk)1$)09u$0%K_eY*!onv=1h>bxwxme(*?EgEzQRsp?PkO7-!IHUUX+yb{# z^d2j$KaN$gOo0r&pUbzUQ+$<-o;(Uf5K4erb zY_WSC)0*wHM)3Wq&J^b(AMI>CiwR3sRXa>o34TgH5 zwB3iCxN-okmpq8)8%Ew(fDMI(x~Iuvk|8(XL3>HOCa3zd67R+3F2(7I&0=V3U*D#u zm$+4~wS~L))`>x+qkz7|Dl8{e?868`)nnd`XaPKIz%?&=8wLf$om$Ggw4S3_XbR~y9$tm`fxr4a$c2Qg+I zBNmiZsG?fzQJPj=6eqknr3n`}{Lca$I8CdvDDCfr7`A1ND3=`&R6bVd6Igy|l-3(u zJ%uiR1SxQ@Sp_Z!wMfePkvkXi-#^lrK3l(l$M}&ZeJhA<{F~>#kU_hEZa78^4&`#q=k>GS!%f8XElk80cVzOVbfulu?l*YR1@EmH&TLqdl@AP~2a zp{_XybU+UT+8ca;69n4hlG|JdJlNxHZg3q`)+4e2eA(x!ZK4eVRmLCQzP}&%e(|cAYD_paD`{eZ5WRF94oeNLcb6)JH{3Oh4U|=NtJlzjmz@TPQ)aRK+TQ@Vc z=BXIUvUd-embQY_2ex~5e3JACZ{xqFvRq+!^a?9)(+o{JJWO$uJM;4tb>*A|X9Cszi?Wmnn8HEntit)vV*O)aR1ZELY4M z(Y_g0r*@_1XtT^>RGXBSvmbLOwwX`DXQB5TU+FcoGAcU~TnDk**TYiyOD~_NX9o%# zxl7IX@FJ|0bC8$k;j1uR{LG6v6Od0R4^QF#)vaLPcy2MNaE|>IvcRe7i+4q&`rNJT zDG})=Qs7{i^3^LRM5Tt)xBDVE_U8}qG(Ljz@?+DGq8+^8JGMslFB$2Hx`muHg9x9lq>2X z);V7WVRVD+C)TWGkEY6t7dsT^ufusE{_ma~Qc^d3Kh%cmlv&wv@@4A1fK<7DQXp00 z<2)a~mdaif85_eL zT}r`6subXe7xi(Oebi&+A>>VEtHYuBTv5gG_byj{#o^v2HyoFbYQOU(EPSLzdsQ9n zHpT<;bKiz30^B~X`4HZ3QD>0byR4Ny+lMGur)RfspdD=azK{8+_IH1!uMf%xBzzDv z!+Mus3a!$&;*xoej)8`>C3|g0&N!GMiSD+TGgQO0Fx`Pc`Ns)wtTBhV4l0UXm`FaZ z5d4|=OMB`$ou9a^eVwFPp-!(U6P0T6`5a(r`=pFRpNp5jf0Ld&zsMEk>-IR|qm;{{ zn!T3cV4m>kmA2}&63vIrVug^gQzbvW7b{n5+~7vwa69Iyo-=a4?^WIq)i87${lUAI z8lypwgvZc5d>E1uD}E*`uh(kocpmmSusBO5z$ z1kM4KpiiPC#%+8*-|`{Ud6%50yDzyev%cQ%T^^$m>T@>(M-t*hq&S&pM*6gK7E~af zQ0mV?cwY@cVd%P-KJ(Ol*KU{m9#N^e!0L^Zx>w6bOmcbyN36rOj#xq_r0CgcUW{6t zYp3_Od|j?6V#wLC_X^@tXOVT^wZG(lQ}hI2qyIh6pvU)NM1mpjCyGC0(`mw|r->`d zMG9OqS}r!|`F+?o;p1e4a1lrN#}&@kJR)o@a5CZLcON>Jikxc+o2GBQMwJnmvtZ#6 zOs?2?&-AH_g_2wc-&zQIsN99x@-NH{XLoXV*v6p!G8F_JGx_tRVk>v@z+*}8L{52Fc#%|O>v89Fhv}}1iit;cdse2j_BlGD1m$s?LlfPC=p!3~ z_ziQ20ap|g`Pr(Y_T!!SSY>NVbt$1IXVQnW%fGJet znr6$BR);;RfZIuZ9@grE-LQ4>sQqPESbSVmDkWWwho8w0Y(C%e@M4w6dRVI(U?EU$ zL8X_p!$*TGYtgFGy1zCHp(G}L7ZvELCQau)QX zH6&{9h+yeK5M%1v9>+6i5T_%g=1j^8?)|+A^{D5%;(_+QYdRdHs(DSp zW<@Ii84t&-cn_Tf|8B70JyL>eWhiV-jhwqYF=-o#jJThCke5HzL2t2$i*uNilXTus z=BlAm$L3W=y7d6I6Z%DaujnNtSx3bMB~P#o5J&fH4#LWMSas2lh@Z*zqku%))%t7B1CrEo{bupH6A zWZ9ill`D67k9ggpqVEA5Qj}}xwU7ZT0oJ|qm@>r@VAP2$BJ+Yg{me*>Tb5_?$;Yeh zkYkzI9cpPg^BaPVi7|JWxuKnGBW}oM0LR%Kvl-*n@*y5j!(ecXUz6(WLamroKgZ(5 zw0`eKm2VmK+b*5=df$fWhO#%1y(utUCEp)LWbG83DX%+T6f!WA<~WceI%)tpb*VwY zRg5(mV2j$~f4xe{y%n5w0UNg@2^iT2zVzkG^KFzztnGy;U!On#{M#t`dF@;WN%}Rb zg~11X40gA3#qdL>LjfIf!$2K6Ia9bU{M#Rwl1_`hIna3}KhyS)`(HHHU z>3;o*>h#>bCS3yanTO_k1$g+=FZH(EmmL$4s<^&u(8E7mD}H%i3%1vBpx zTXpSm$yWd#`eb5znu>_4YdYlYFdYx=cb!(o`DT)PgJzj^Iq9djI^EK!>!-5KZY+Xc zl^k=D5F zU`6@In71eAAsc^3;jPz%AHTf{_^IlZdOq|zb@889nIi%M6l(3_9RhtPU`&xR;74Cz z^u9zf2STmqypsx@&32dK+2 z`*O@&(}BLm9d~&3p#PdXHd6(|=n6TmK*1u0XJ2ICHvO<^@Z>`B-PaHAa=lDCyjJpD zFaA@fZh8CFH;SS5084&&C%C0bi%8Y%v_cym1E&1T;g zFvG0tZwb@=_z1_t?_i7NIl((?!-xh3xc`P1v1UQ%0x|Z*+kMq;>B4?8w8b^;F9V_1 zxsFKp9@2}qiRZLiu9u>I{ip)ju92_S9y`ko>Ahb5sZ@|P7X@YA^VL-Rl7E+6qTIzR z8W_}xhV_{9@MkngESP@b(5Jr9GG{3-ZN>MFY@c!8gb(i={4w5uY=*}LT8^neAe|cq zhld(HyVrDZ%1B9~N%hLi3x{YQSHbr$c~7jT2MoQPURSb*$x?svh-yCR$R@#>8&XYC z4P^^Whg43J>phdj$B0|MCiWo{2K**`XJI?bnf)6(KUZ>lYla($zw%Oa(6&&N74mm6 zl(#=D2qq6m^LfAFdwz?6;P1b=sQx6;?MKTd27{F_V|&E0y_YDV~t zsTWQBr8#Vg%(S4+MeV%EG07K^I*RsvLr>3fUy5IMgSUi*|9o9V2+g6UzYNp08&a3I zTi2cnL2F0Kjr{`P6BK z=-DfVM{H_w4+nVoH{{I_M;R97i17gQqcUU0npKiGqgc?A6gb4;EC=ZK+h3luu^-o+} zQQ4ggs2(iC%>X7a?rM*8^SMLdIod=R1AcxUwy$bFU)Oi))J0uDz8)t$^LD>+x0tc! zA{&6IO{U?a{fHLt6~Iaj@WN-V%>i+0___kp{vW zZ?x=-Yxvx)mY=@3>u;;Y2FpsD2$nUgxuVS`KAFDsY{9w})wy@D&g?#w(NDi&JF1f^ zc7OpcvH62RfE!+2@o}V`0r;i! zB8Aqq@>NwIID~J}O?AVXj@P{W$$>olT|kJtFl>KV8kg&C^1&Y9W2xJ;fjT^ms6_(I zuGcv1$9BH1<9O|@rZnyUqNdiI5WH!xywu@VB^U51SSho?_rMFIa*#OUohs;`i}`tHvIX%7I)5u55} zS1L)L>@L~bU6p<7p)RQ#=qcv#@ITE9w_IX_h!xq?^cR!UQBLjD;{?!!Y~-z9r}j9; z$Ilo~0{`-cI09c!J>cR3f_35M86FNq1~JaXH}SOQN8-$*U(X4MNhx`c77jEmhK8@f4(?C!WXJ~{6fJNqx8Q$ug)-J z7ElS>!)Y>QX2!d#xo;zl1=P6=aw)83@aCt-29{E(1Qq!ST7&(1!C}BZ!z$4SP8=4O zT1zbB;cuu3Y^cF>Pu^+Q-kbJ%B0$|NPTy=n^ThJ^=n?8AmZf0~4N)7?J*JANDYHjT zttguQppZ&tPF1=s+Iq=3%2LH%0-j#}G6Q*xF^4<}o^K*sTCVCYjy;mglK;P)+0B^T z!bYtzG})>Wwt`yq?qNLAyNpU#>n;{MG4j{Hrs9^_eA^{l2826aLO^tJ66|VHqCUjI zA47fU8z)J*yF6b%c^r_Ix_C)$c6Is$*NBy>^J~*VzyXhNS%@~)w zSlmE0CZ-iI!CN6e^sejdJ@A}Cc3#z5L~B(dS?HNtL8K{~p_fCqKccQxVKpW@7VZrVSrob7BTY)xd$n&{~41wA`vQ*-*Zxf>AV zX$?ZfKMsc2xsBl{@ppXeQ9166rto1JY2uxe_}&`U-ql62fN&~~Il4@}8tr*!JLE-V zNN}1J1`Hy;@bPJU5Tpcm?sSdlOV|`fX_#?Q?ff8`hi<_PHxs>?9IqS@6E(iRh_QO@ zLBk7vLAmo5nXWWw=6Iyo;h>>nO?f!pl>_QgRqIw@(+xD%a>zw*BsMbgQqw|6$h~F? z@n7;jeJh8WpW$}J(Bsz61R1Nbgh(ygE!K4Cl%f@1{WNKj-){}_aLt~Bd7qn0y4T~Tx}MM4@}0j$Lspjefv%NXf4v{1q?YL_ z{~(L|V60)^8xZJ-fyafW0U=&lgSDk$&as39!F=H7ijT!*fjaA~YD59u8<)vu}L~*1YN+`kLR+4{W%je7LyGmoRc-9o3b#;)bI{t-V>d zs)hLdoX(zRWS_3w8u9$07^qO#^4rK$X2$fCg|*~#0Xugt8U{_&2}b0 z6S1T`hiWSS>GlgF*D#A^ZHRCG{FsuWgBynjI`DuJ-3&`!K)ZvB zE(pmL7fo378(s|P4#He0%bQ(L3qi~SNtr3thaipEh*tQL<|&%uh7iTL>^5J!m8E8R_)?e7l9(9{vvKdZT0XQ9P*c$KdyO2$1d>rvuu*N&Qv$7 z`RY3igepR-4vh#$)+4s@oOSZc@9pBR(v!&%T~?yZz{Razm0sbX*s{rr6=>Z45r#P` z(DEAw0v%OEQwU@Sf*LZ%_8=%_D8bNQ86AwBMBS})Nv)e*_l0Jd!wfZ8Rf0F~Wu^ot zkb<{E12ONIEr~;U#5JNQ;e*2ZovymL_|k&R`@?xilxLg>Kq)( zfp`VKo!8;CNz2bK|Ead)*Uefz%qmVl@4%W^#HhD*)jyg}QvzkwMe#*ZD~9r5UNuI( zd-3h522#{^ND=OBwF|zVr)~ZF+xC!;t{JhnBD-kmAE$!Wb1~f+Bcd*pC~AJJb`M7O9h?buW2-BchtywqMhd zg^p$#pEpywWUjKNjYi}2-xjFs8L#VJ-a|dc9c8(4vyx-!2Uyz^e@X2#O`3hZ4%Q7} zlClEaF9(TN-U^oM5J(m=% zMOy=6sHL~$T;E0U=yVz5lGv4b?05%~F-o9Y3E&lHtI$xqv#%n zBPz@n((aX^ybDh!*<@!x=#k^Oc_QmM{di9?h(GFGvzGXu>H<~#!p1U8Rt=3 ziia^w^NFWKVwAB1u&(MB?MDR5(>XDoTZvKOLN$p}(Cd8yE<2qKc{iw}F5b3nbpO~Y zX8JVN1rIqst_Y8%&qgEVZsNp0l-1@k7S;)bD*%}^;KJxBK}FC)V|&b6!!)_G*o(jY zlGpHGLi}7jB7JcW_iXe9eLzmxKYXYxCFl9b@Py~BcSU9LrnD#66{+Eu(?Tl6<=j!! z{x&{Ui9Jh|ixIklm-0^ErxFBgkqr%XCpId(9UGmiRmKzEY+N2)D11)Iwbr>_e_Nq} zn6uab*xKs8gcFTt^hok+P-&*G)kgKYbDDE?%4kBshIDVC!oavCN*p~s)4CRq+%8+K zDRt&P*yxZ+kb=kA+1b%;+T9Etj|=e!$&F?8Ws15&>=f%%RXpEv? zs%7pkAb=mEw~j+Z6zhE_R#%*--DfAjAs-u7$etMD)s+_!`yN)(;DkiTiU#%cjqyj@ zGONOywMRGV*YDfyZfN8#pn!l)cAE{99ea_wV$g0rGCv%){IWae^d3i^UK97Rtj+6( z7~_kTv6VJ|R7gqcSzg1HJfpDdA&FrKSnUXUqy(@7TG^-fl;|dMenWO)dGN>6^mO)q z=D+vmQgO=#&y!>K8=b1z0L~P=Nv&-{w>mf@J5JlGYEJ_!rPmcbReHu=hDH-3;9$n25;g7pw{Z#UqtjHH3L@^w);^B={9_>1G&V6i8c z79Ik*d5`=bBiUvOpMurSS7fW^rCSBGh-Prx0D_>Hbc`%q+yKcsDa22@55UKF>Mne6 z=3|2%NMIR-W;Uzn$|A=8RKR|T)ye0E-AwzAUkSc&@8K)Aohs+@ihQ?eIA}MAv!J5u>AQJ7bT{P_lTuLJwWT!!Q#Es-wt)>5Yt?v4z7(wpx;+Gh z$H(7dG33*L(9yb|`MJi-N(B=f>ocz8u1p2O%Tagg^-;vrW+yf?SjGBxRW?%Vq4C&z2-DM#NkuA^qL~ zhKNcGVEsozrPw#j5RnVLMW7wdYe8qb#H^_QSsv7Pu-63p!zo2>?RFVKjyQ^tT zEL=H}9?VG6l=;ZrH*Fw8&~oYZ8%~ZawZ6WQ>$}#q{~5YaoZy4#y=d0nLY%=Dj9fj9 znTh=$Lr;twp)M`cg9H>$fVQO7(4)Pc7<5AlahLK@b=WMbi4VR_yf;*^G}_a^UwJH7 zeZf3mGXyrZpW_DDp1Or2b_UyErw!(d0$;B@ zeCI~buXa@@-DtIZxzaihxMH6gnt(i@N(K!ZPf*U-FyghcAA{!GhYWc_R4;~D7R_fq zGB|Mh0eol0OpM12&G@GAWT!B_8XQoS>JLODFHv+~c@-8=^Za>PF?z)gRo8!#kBkw7CgoR4x+`Suokxamk20aWI2ec{`9xXXq5ZFNyBPrX%miwSGzbSY_7l>DuYvZu%dd~Bu+plphQ>)ZXEU=yTjicV< zb95IZTKUXi!YvNnd9x5i)pYid0DfJ}Vrv7<}F!vEL3{QnxpUnO3e z2vGnXdClHyy;-IAb!3Ndc6&S!8Fl_!9KBNiJRbG_8xPV%vg3{oFIW48MiWdW$Z^0& z(s{82$wSiU&32wP%Ax#2=3dcRu)Qgy%S*1pGm7jem)7l!rl7I}8ptw^1U_JD9 zRB$1wE>6A}VsIIWXSm}xkQsyJipl?-!dIaqkWSQ;FyEHqAqFg(X?k;=wB8n3hF4qH z4|mH55&MQ0ixs=6Sb?@rX(1Hk58Vo~xq)azo}0ZYTO7dST*ort{#)-$<-W&WTrmKL zc3s}vJnGnDKTB}PT0LFOi21%zp<&Rdb0k<`yne-C0B4EIcw{{?%ebQ^ui25Vmq>Ox z=>s{Q%lSt6%UwbF0l%S|Sh3RZWmeZsv?r+Uj;;&=9WIJN4TcRFK_)T6Rn#-Q!Uq^G zKq#0@&hhtR?S_m1pVjD(&hd9cl$JI>J8Y3(YN?8WN?NLf_}U;Cz@+$RTwx9NEVO5g zTm|CJpyA6%TT;}c0{3ytnq95Y&@^+?1%WE|cDd>9R?>Z#QXzAz;&H-z30mHdoZbrU zXBU~2c{_}K!zVrvhE-3ss!D9TAkc*bG&(r)A(R6j2fC5Cu~ZE7F*p~nw)Dc>beGT% zj=py&cL#%%i_0h~60K(SP+dcr9b_x#Q2i@an50v-{tI7;l()gS&%fXMvs!6ip}Q#s zqlty&w4&H&{MOw*Q+;_Oc{`ph$jw6ZO$xN;i>cbM+h3)wv4^^G2s)#W7*VKte5oHS zgAK7EVhl_8?#GwY?RP?pxG1w#bZD+SLot2JuLLr-F#vluNGf(I72?BI-hVYSZS;LY zRM#{v9;QLMe*5s~X;D=sb%7-V!+Y4d;)OB2F$Y$fB zg1B*h>ZiZXns{6YKGR%Sy863GBgku1{f$}ko@{H$FJ^XrTw-<)Stgm_XkJd0e&j_cMsEv!p zmf~f4StycS)=ZihSK0bqq7ifkFU!Uc{-w=Sw4*;i=-D1)*VY*O=}FH(KS_>9x|<{* zsS>X&mZagSzyZ~DJL;6ZydQLJ6=2;RAmecLjcYG7EA{|(eM8N>_Rs`v`|g1yDjV z=HY=X5}SI4v(MD>9rk0W7XUxy5n-=g?|TnE%i;m0y7xzJ&PmRzi3;(JcWA^$0;|*t z@m$5EyE;?+9H_$LS6~c_9#_gGP=FAy5v=!1(#);>ODtJ0<$VPd^2-_Xkz7! zr^dV9>b{GX78%j|yfVv0bq622VhmDq8jp>nCQJs;E=uCUDYTf3+)IZ6 z3ayzw{rgzHw+|FU&}xt5&RR7uY(#CjPn7RWS{2GZd!#Jk;Ik<`xy|gK)DIUk3-S1L zshDq4ugfZZAYSvfeH|jWxosIsTwZvNCw?`ibQ!#`VXh5z0(HNjSSBq(!IueCj@5mb zs0jqEfhG;Iulb>`+t8TK4mHzV#Wbvy4`@B13AR&?rEO4@wtl1SiZgn8D2hQXt(Ww0 z>}_I947^v7v5{Lfrb_^Vm!+crX$3NTy)-@dH7)mW$)rR*Ds~@7ja(X^1KcPPGyImH zZ0hhda%<;u(42vh;-C<(>wu^JX@|AB5StjHB^vy7ERlnA}muMRZsu%q0DpNh|wB#P$Uxu)Sjxe@5pD(vb7 z0w7I_95TZMj~MlBq!y}5$oVmWG}!?ZF42Lg+fUNp!!CyQd2B`#SY|^(>wUve#lb2J zy*|WasyS#K57xHdvt+9$IW>yXaMgfvKlg89So^*et7>SX^@@v?J5%!QR|GW;T#g5& zFKFQk;oMhwH#wdF1@u4JSj>8J)NFM6cV=X3$p}KBKeD2|f|Rng4yOvOf2lXfW%Re| zhDme_vv8)kks!AN9g2bRI=B#Gv?Rvg%G#>S3@8lW)dzjI#72%NqWAD2cqBASH7O6j z@x)aFFqDmnux?Iyr=J;B!BV-k$QpMXsN3l%53WxgKFAue=d~CBawo!t73)-pNL@zv zK;CdQ0~pCSj;h4dXB*z75T%lP)Y=Drz{7IxKqxh|Th!F8T_?%RH-J>f`cxkDzEQG&l#vk^+f0%>QKHAU)M`f5m%c#l3#dgCZ~y4^|@$T zE*Xh;Jv8p6T9&DvRaq01TGl+;oc|{eKXjGnhVp9?KKR<9eb@H5>DD~tvP5t?ozZM3{+cML(J+by5^(YP z>P4#Oide^O_y6alxc4=mG@qdXB&*=M;}a4~zVnYo%*?X~vo>!w48X&tikbpsm;SxA znEqw_1lKdPTYn^gadMS>Pqd>&4Z-;*O%`@fr3bi^Ka&mpk#{TV)S((a)o@^c)PheqEXTN z5=vI87_le`*vRF;)qB(?)NOHQ+Y457Df)VG`H$fcqq`7XEnzR#$_)dwObpe`xxYnp z4~NL(&_%@r=3geg0&G{fKASN@!;rVGLjqVJ6=ah?zOPlyTe#okNL!Jo5oYm-QM z@_G@T;`2Jy4&E0gyUwXMojFk_;6+W{h^ZtjRslM!!PbcRYYz8}Vqbc$&6w*;EWwh_PNC?T7iqBvV($&kFVha^Zjf>)ZS33-bOEWd z=KIrkYY4nO;DKjOJF`AD1kZ9|8YokwILt-GvJplE#w%JohwtK$%$*war6+4PNusTc zjP0lDlFG+T^JFYwx9;V4)s$=<(+^tl`(m%O+3! z;?vMMf+W(;VLd;7=`Gihy|Ix8=kL_;aQ&TOyg$Ru@TWb)I2mzMtL0S$J!CuHplEsH ze8^*R>oz;jx~ein44)ef@xKQrTAdlT_kOmX%XhmnlWK8h2P42tI)pIFC?W1suh4I; zBBH6ay$=2JB0uYd)d3^My%iju(mX}W+-cHvXnNFDCfg?x8@bB!-Smp1j=OG2)@dvW zApgV&1I5Bc_BL*WkS26JHWv`{jZyW(CGmUI2^%ya^iN1@<9KO5b{)B8KFR;5#J&=O_3YX9I`|Pwe7Wr{H zPTi}0>=PNyD8d;^?j0BjI;%$C*agYB5v}SEb@lB|BP%2Cv{WSeyM)^NB0tOh2gQO) zge%{M>sqA))Pu98Ito~KB#OlDGZnxk3yLB;G6DyjxZ2N`ef;8=zrB>HFS)n+KUT>2 zwvnq>^GW{rSC#fTxoW=bw$H^5j0TS_ME!(Ab0zd`J~+6&W3@)1TZ$E%h@k&!#>1pd?XYx*;ZXea$$-ptg6#RHg1awBkgavWn~5*hmw$4Ff!D z+qX2`Jj_zZPMsKKn9~}7LXc1@?LnSzFWJ=T`CRGY3JoA0{c%4-V7TMx#;2jQf-|3= zVk1Ep3Ia`SZOJX;MkU`h)kSJHfve&asIt~9?Elu~XanUnP6y5wH#hI|>t z(p=WqXAh-Pj*4J_UQOT3K0k%$2$*l`AjmqFDa41+1_qu~j}rY*_O0d~gweV-y1_LZ z4^Wk^5E!kA*eBC4bRX~J4Mfz|{OWJJJq(F5m6iAE#zO&BuhhcMU)g%I{q*8Zzw3=z z4JsWeCXo|52oq#S*YQVIj`D0*n@0i*v0A!g8IYxWb?Lp|eP8J7Ia2TQ_|bzZaPYdU zl0=`elEm-}4r$I-Q?K=i@NRJ+s2MkDH5pK@hN;7;2h8OlFf zmH19sO6kx=WahsI*CGBKN9629N57ovh;aJ?KI}w*-Cj=vNhidWTTZ{09?*UsMd;+h z%l~7R{Xb`gMn&i(*V@?$igM&75;8mBEXU% z6`(Xf>;*`nkN?nY8UapeUEkVfyWqR+D3+*l_%f-;couj^<5Fdb(OSn{ANnq>A4@MIh8CjftVu}aKtr2vdz zy6>3>A9-Y^OorEw{-1&z}WcvAW2NiD+pt;~+g zgy02nAwJUO&m7VfiujOR+MP^uyedyVaZ7oq*3SpLJLf`Xe3m9-^s;S5bj?F5aybzT>q&ozR>P$K6cq- zdJxf=7zTsog>*}IV zh3Sj(MyTuA4#kt8tyUqv%kfXJ7c0wwP)#ytegJo~>6aEeW8bGZ{H!8;>B?r&G>xW~ zrZtTB+X2!Q@bp!`I>Dc=8$!i`M?lOeVcv%AFDo-r>Ok6kLd#{K9ggq8M%A00gt-Q_ zARd-i!r7$M>PA$* z>DYfI$Y*DH{b8kV5qoKVWg#DC_IQWAACPKqLUiWGuh*sL3cxZiP4sO1Eh+64Y12y> zQ5!nU`3#AzEJX%4O1n5^00kFdNY#RKGjyI9JXG+qzPkCB;Xz@iVTx*<>H6P$vLHTQ zF>x>f$}&(53ScCeh?Ah|2J^*QE=}pn`wec4{8gQS`NtM6VXZ%4x4lz8Zr8G!*oB8b z98f3d{`c`Y&iI;Xyr+55EO^68vE<*p=uySei~jKJ6~9Vr$vo^I+_)PXKHi}@@y4mE zc-oQ8FBL>dJ;lTaz4(W7?>iy_fZ~r}j~P%M2fED56>?|0I{#yS`{;N_LEcG&{%@=* z_0clE`#CQ5T>z{6&`~{B6sd8GWH%eZKOBRqgg0-xNj4FG8*luuQ~LKY%|%`PWinG9 zHl1++`ZZi2{c>zBl9BAf>sLBd61%N3+E?iiZ=+e34$GkoY0!!7fDDEEPz)pXeQ! z1v+xBJQ-IYUtPEMwTYLf4P>HOKLRo`F@`=9BR;9`)M+u}sC?4==x!T7$&ViAFJ271 z^EbP`pn~xb(wHP^V!`a^)vT=KJ}i%L76`rNco}0o?K8{T|v1? zU4Lj#qXX{;3gAfGuYl6Fa^=rC#n;@Tr1dwQ$$(@5=Zeyopq{8MW86c`q(+g}y?ycj zJRG)$Z*|Eyqnh#~jH}okDw0s|ojEHK9B%ejOX-kJ;1@XLnDT zmUu6!6MBvPgh3C7YlH;;!2ScB_O0`j*&93RlH<^1Uc!^7PNIyT*;BYQW#(M{ghs}r zM79cw*tA7h`##hRaVFgb%9xzr|1C{`X8lV(!nrS!=)!gdfEr%7_CGX*V)0)zWlz<= z#cWw5y?iTacP-ibIzQk$ZNr5Q7eGf6{y%ld@!8UURUR*!gKxk6HzKY0KZ+PBfBtm! zAD#VQI{`HDKY@qi0Z`lpW_fK{!|+$tRySm)m%S7}_AV(~{-4bFC(LAydY1g32l(gO zr{C}L=s&S=2KApMyMI;hQslNA#AVe#)c|Gh`5eD%f{B?5Adp#%>52DCGyOoHX9i`S zMHKv?dXn{XZD2p>m-~(`AM?{8bNygZZ}&YV5gV1N%cf;RX8GH*D-VccNj6<5qxjr; zj*_sAPr!>}#k6tA`ybwAr1d~-CM$POs7V{Re%ogcC?FMlTQ~jQzv)~ApIB{|O?6%6 z*lSCkJ8E~vk9mh~jzvVO+1A~pIoH$IkJ{nN$%WMqNiq4DBt`j*GQ9Ih4c3<>#6;k? zj_0~-W~|7-k)Ly@6ICq-C<3zUw9I(!l%aWJL`W4~Z*89CcFi$ywO1Z&4+3S(nH{|? zgG%CqwBNG@6=x$#b~++ad2XN<&J!<(2QoPuMbLd>F6P%m*}|W^rUlt(A{nzp5WGhzjt}lC;0Joa)H}#w=y zmyW*yH^D2M+^V1c(?S<$FE+(xR+l>~$O9e)08yFucb~ue^Zfk^kO+;PXRNc^VbU+h zvmOWAPx#G;O_O$0v_P`_BD%~a$9=rOoZ6~1YafmTr zO)EzDixmT}QMD=d#@NDrO_gryHP~UVmlFpc3}x3qW#;Jajb*;u8Zn&5azghvnl;)h zfmVSozPlv1im!7~kOJhm>u^?E(f>&!-*;w+PGi?lTc57tpc(V9Tz9j~2e)*?_UD*6 zjC>C5zATg!DQ19MP87^Qto_|`APcn^Qo5JaqQV6N5hV#xv1*^*zuZ)t>)&f#W)sG@ zB@kA(G^(LM%}a@Tb@z8GXcEAGGG3fSPmmLFbhphNa)x{ zo*}7sLeAaoT5pnqi|#b!FVh#zr1Ytq>tKN*V&J0hCZihM=+>(^#_8Zg$XJ1tn^6Oh zfT)9#zW6)knkvXjqT>KPPw~W<>5n`M_1)6L`Kp^yN;sflhy2{rLxM`+UcYiC?$dgq zW0ZHbO~~<`u@~GCgUhOpS)`4I+Kge^_SzPd)(FEs?XK58c;pUsV#JJ!r21r5r})SH zQ&=LLf|Ryw)wayGnZM!4brus7_ea%UbwL=tcdi!eC++|I)Pyt>!+EO2xy0SB_t9li zBHwY}XWKG-O1)a+r;v_W6Zgx(sv&la2jQArSYOQ)&Rf(bCe)XZ3JW{mfUsw-GMDvm zOvmst{HfRs6a*3v!<&}h?pmak(l=yot&jK-#xjl=QZo@_`e6^?5FwUUKjQ|O+nJF% z#{g$~hVl7jBE<#dDU58f-id1LuXmbp^3}`^{$;T7topXSDY60d>@@p-F-YhEO%tD;A|VY=n0x>>O0WwjsZ#iQ8Z<;%!x+8hVh z*OeU|NxEbM3_4}h@y#PE@n;pjxlG(9!}^m#{WKc~>R-9}3~3$b1Q9<16`==ek6&>7 zb;EY&mP74k%K@fQcS9nCz7i0q3%;;GmYhPN{wkiNu&`hw>9QXDanQBE#r8Gdysj{T z4OwyLfrgS0{`P^azqzmdxO^hpeO%>>qi)#dd`jlZJs-($AtIchzKaXj5R)5jJ_`23 zT%xA)7Uu#dgegPO02a%H;8hGiEGy)l_AV~x8l)Sz!+W=UI&Z+oG zQfFZjhpc@L@`D-;xZcn`-Iq6N53kl^^T=Hu3D+_+Dzsy^5W`&j+c>3~lAYSac9!D7 z74)6+66Ys-S+^N~5O}=p?g@mV0XBjQkqKj^OY`P z${&KyaDYa#0G4_$IErvzyO(P9Hto+L=h(g2oIE}Y^Q7}9oN#$sC$d{O!N2}0y<^KJ zKFcB7Q1)&BFqVDdJKKcWnS#ThvwCVzAvKyp>1L?)7A_{N&3uzB56f_+6Zav-?A~g1 z+B>+H)2D(xky}sg=Ar&tA6=0~tG++-lx6NHnFZdx3?LGClM;ero7b!^z*NX{Ul~lC zBP3YaoZAK0KkxwG7`Je7=dfWz*t5TZItcGbmX5sOtc7LUf9SBUP^*~6&-)Pg`K z5(b5vq+EJ`ecj+xi$V(s`#v=~$zD2^=qY4dMkERE64L^DStcHZ@o2g46jRO#PpVh` z)BKIOQ*}~o4X);&+#M(eh`m^qHkWa#zk`VP{{de;IlgjR0$IR7(vu-&*vM(- zO@)ikxJSeQE%f^m)xss+Os!wy5bwfnK_ex!+Uww7!NEN%ny%%+ZwSH9CBbpVxHol#GiEs?NJ#K)pg!NHbMI zhEsg4i6c-s>*xYhy~kyD{H1@5j{~de zuBR6r6X#a(Rlv)!_oiA_@s~UA!{7GH+FRQg`5Pvnd&UX$2HG%vo8G< z51Hjf-&>9mWHpDp@yp#jFYHc#Qc9tyG8@O1G3Y)dAa3v;fR%Y|&!zv@Zl6|Z-@Fmc zzJ$B(&tvYK;S^6|BLu*U%OVpE%aWMQo)Gdz%lMw~;t49BDoj-o(PiLYrUCaH`phb{9ntoT+xvUqAO?Do48<_dmrO$T#ueB=oY1D=HvNea zMh6DAI#+Qw8%>t4nxrR)ft+y^@-gD=qE&|{$hGgCIh(Tq!Tm9#+hJo+;x$-zQ2392OK0m{GtC;SbO+Or+zlQk9h# z?1uorAgA1uE%S0Q0zKGyd8h*1io;|*9JNRcyB1ln!937T&mXhc{HY5u-`jOIOwiAv zq4>T7EdcQ4%~g}Mwv<2S+SqJK;Y1E|+r;c$LAW&vdyYuZ4o@;3 zqe4ce6mOS_a9*=|v!A`eZhKma6&OLDxCz~5ecva?B!dz)0t*Lc(dp&g<-PZ{@W;>K zv(Y~{qG1DH;$EF`hV4p4oy0m$GcU|%>w>yG1nlD1yZ~h@zZB*pZdos4{*5Tc9t%;= zCl1?}I=&79`XeMGFHCZtqR6H@NKsW~jap`bKV4LwKhK(=5Ns67E(N<$Ck)5SExCUi zqXJMN8J?^Xj}Kk@4akO;VQ+Kmidiw-hKf(%%1Sp!x04seN|ECf4+l9B38uzJbm?jG zyQcUaqsVU9O`+Abou|jIQ3&fXeL{!z?Mp~b;Uf#2KnIk>r!4ZlF(6r znlT?blNQ#PjBq^5Ff5l^?*<@V3DogCP;25%j*A<^E zh@%}HX8Y9vT+8sO`iGiTW_IM~z5!>eHZJ@q)fqi{-EG5la8Zx?D(=QKW5byv#jaNm zm4YO?C+>?sFii>)AOWX+1Kiz&i-?JW8eyhJZL!7q_H<jnpg zRZRBTxf{_{RX{-RNm}mm9}S@1v%&J^HTTwRrAnLZ6RgjPr1;Z^lX*UGuXS7U`Nwb- zNzoH`V>=g5)NjW8-Rw=x+0GwFenXkGLDu#2;M~EoW|O-D7KJR@YqmEImb%DQrIgiX z255YgeZ5}PRAac*rTvsLPv`mp5*7j|8Ps16x5K33i<^4l+^LO$Uw&&gR)O$7I|k@r z-zuu^ZqoNC-R_J4{y9x}OPSLiMGbbO;apU~*5*7v$yb=YKpEJpkO@*E!aG#mlI`1u zLW5slT=Z1v226Nbx_ReK5vTCk<*>6oR@3kEb;51jW%ZfuqrzWTGvrYG*+wCGGu&0+q*O3m1SatfkF?EO#^Wrx-yy?Z8d{)L3 z=tta7dHZ+gwSffh#h4D}R12~$8!bCLjjV$Mu3(r{$He{qTvGfB*j(~dWm+(Fnez5b z>~^Z$cVjS=v3CIIU{`x8`p2+^`Bz*#8z`U;_NC@*slJF3 zub>NTjxW3Ap0Z5wijXSm^3tB08+5MP;#Rgj)2cmk3)@s(Z2?qgG1xV<-!1G`nAGVu zM(NL_{SUfgA$!E{mTp4e$@S7Z-r)>`Ag(VO&xce2J$IkD!fmmRXO_@1uKz@^NF715 zEv-^PRk`VkZ-RU7aj`SyI69kKyx$!r2y2mAKDnjO@#0f*_t&Gt*X;NTyvc=P>TSWb zrgka15^GH6wpf2S<98@N1<#Ep_0EJz!Z8Fqv5Y>&{$N0n2XGVXcjB5NKShch(J^LEXVlr?y3n|FLQ0=^QnxnW_>|%tp6_n7#SYP0{r(PyT3hQG2TZQm z#V7k0kN?@oY*l2=OjiXez+~2w0*)*~?m`mX&XgTaKlNvbK+lBq9+}gDFK=V0-rO_n zx16%L{2LddFU<>-|>-AqCHfz}T@*hUWkv|>n@LR7~ky{Cj*6ItrX*K zY<3|yVh6BNfBJ&DM7U0-IsDSbcgopE;C3&F?`}}E-ihCirolsg@byx&wxZ5aJH9Po z6_Cwdc{k4r=5!M-c0@j6?OIWd9gOgBc9c2(zR=Eb3!mQdHbRX^IzDxvc~z1^TubU( zjNwQqkdX_CAMER<4gi>yUmIqUBb=WM}H0=fv6rOQeR14Vva0|5A0CwO>v6UoGk3%*9PzCEhDA4I5#BU#Hx` zM1HAl2h+e802mE`4@Fdv%GsoDZ^J6l$hN>PDCb=3Uu^Op)?ApKKS5bx*&%BshVZVz}DPZwq5)Z9ywk)fEaCaswH7QdN0Pd9RQ~B9y3vC1Oi-jl3V5VE}-oMYZG)hRgjgY54#imB&1$cJ{ z5}Kf@S-A8eh`M`&4Izd#*`TVGt=j8(4+l7#)!=x_96$C3CkMZe@QQ)5kl44cItkxz zIUuZyMY#(pBbSkLo~aw&}Vn<&{sR0tZT`O3m< z1O%y0BWx1N9To2u92D!rVT^jkh-2_aW)(YaGR}}cKRYXCO0NUfj*$hOruF-N zyf~h;?HRtF?%8u2NA|iU0JD986a}AsYq>u}VHz9>NNgo1Oc@=O_h{2c+n4+Rag{Z& zYTcL4r{0senTZ0@P5A8P;R?zl3yxEHja^pX3xY^<-=s-3b_|q~<^l&?{z^VgN-sY;e zpHx>1x!}5g$BerQ$B!9(Unhh0b2i>7pSBA&7rksR zgh^?d9_#s%xpI%HY{;~{<`Kbl%rcdaq)S;NPY%~xsC%!H#oGv^`Y-D%g-s-x08oec z_;V;MxN^^lJ8MignH3_N-9Kq?Y^CJy9MhS3^L0oq^qqxSU=yrh)3RO_2Q*9Ld+U~R zbZdx}VQZDGJl-AE+&U1W*;*JGCxvJ@-@mpys5pbX2G;NcxazaQsG`bZ=n&{10Ot0u zFIUsWo~6O1WET%bzoI_4)V}C!?yIUaRNJ*Lq4bXhj_Rn6VD)uKM6nKHm-YaPr<``a zwUnaEp~~xQlZ`v@_+rBpVNJ85dz!PPigHzaS>?6@+NI;o8)5MupvxxC`0oL@XN?=|_-9M7? zjBYnP8(-ZYyUpM-%vKD#6Y!DW@5H~CBUOb|W94z`eREL7CHnGGSiVsU$5^c3i)z+E z`0ri}FV^<*0Z+oOhEZqb4UUAZ7AgQDv_32PtAnq+?oQe!C<$u0sCCF4R3^s@XkW@=kT^y)IlfU_JPgiI--n%~lR9`uvih-)!3ck(qe5!^4r5SnxAb4pOB=WQSttlf_6Xp2b zjTjIjx|g6bRWZLXGwbp(S^s#2$07|Z%&6&4P*7P>-73AAps%@t!t!LzTaYkW@1`sk?+E}4`0h>rL8^mIrn;|ynAOr(}R%% zs;!koI3}>UxQI04=+PIban-4GhSK8{u4sFz7G5HxB+cy}=qepW4&B6~PwyNxPD=eF z(*OapO;Hy&sk}fjAEYpfZwd*qt|ygHv}B$mftG2ky^b9zA>>o`vX!^Z$;{6u@PpSU znqTA8kG~jua9wxkX53!hiP+$b4Q!c%c-s1}atdFV&D^umT){@p5EzPkD<;#0oR5b| z0{;Q_RZZRqdzOG{xZr5G^N@=Zpgyw_q_@c{Lj{?eo@E;!i#e?tY2UjvRQo1Wg z%CX)Ea1#*=tMn=yn_s%*MPH#?b#d-DZ8W^H-}fmdfBBfN0LL74Z0<|sS-!!z6q_$R z#PbN%V5-URPUAWy%ag%sR&J=O2W@9L0l!~J+7{WnE#4-^4e=X)kFx&~Fa^H92~@U6 zG;*#YP<@m^6%6U%>8tj#n0ng|41y0$LP4mBs1T*qKBH6K7msNi_ zu~m{2@c;L#{pHRooikJQWtr~H!m!RdzDWJg!q8UtH~D8Wx8dzeDz0*`ODxMrU|Ad) zo^$N|mCN{_S1Nhve6;8a*Fa$IY6LU^klp{Rft*Hpr{UrC0r#g36^f{x?}b)Kqpn=+ zV+{&IQu9|_1?u);&E2JhYvCx=P#rwzOxm{(oAZq1G7F5ZTIbaJw zx8OGb`;jzn#+_-bY{>v!A5l53X)@k>nWiKe8|+ zEY+X6(Py?402#oK>(A$LM}*g{e<#oTRp#w3D^~UUBP!-}BJU5duZ`mB!It|MesuA_ zVEcK4Wun+V4F`U8Jf|!(OP0|G7Zpjo7;JOUNmF)p4rMnn8DK@Yk zLgmEayOTHdJ)6z!I}Wuh`s~HuZ@!Bh~L<{zZ@72nS08k2YudB5)=m$l3$PkeK+baW>6#R7F7gz=RfTWib z{f^-uLBNp^FA&7vA#AmN?VHud7rzY2KXbKzFdN%YvCA$rE*b1L4Cb38YSiv?x&B+4 z-xd{kS!I}%=W1wE{TEm6NZ=z~VC9WMiZ9uI2vd}!bCm%EJB__L2+pkD0km8&s0EVo z9WoatXIGz6=i!fdt?JSjh?y#A4%Je2W}+r1%#3$rr@qrYC6HH-^Gvk9p$)0$c_8rkcO?h#BZR3s_f)qTr1Ld5QqgiZ@5@NF{(^g%0**53hcA4nDu1_v6MT-Fql@aA(){IKpny~-pOlbW)-%VVyXBzudS$e)K~tWN+) zjfUkY9^p?_9iOU#NgAm+#&Z$y*yZ(B*YL!%{p#C^^aZE(nr%F*(f=w-E0S@VKWDI# ziL<3q3U~R*I3}=-!KRn?q`t45(8g3GS*jn?;g)imh6ec!gqIkENur{GM40+deeO_g zLlbN?Srd7-L3_r&Ke~R1m{JpI_z@Vazvh_tHDc<(IeAwe4ab2nk2K|l_L(pKgpN{v zUT=|(`K>UC_3r%nx)S%xv}P~uJa4=nDG>i<9MY^miUJK8GZndEFR?awLp9k-s0y_C zSNA~T*-yZ@VzG1b!S;<*<{Biv#uc}@Tly)O>%3)##be&;F-OlYG<4ZOho&O>Gk9)8 zUY&9U)ZNTnljG?<)we6p9Xb>6E6K+#vd8Y2G?XU-2E$t_aC~Ld?frw*{PU6)SDG&p zjm5PrAm*|juOdEWRK@(~V#2%BL|o_}a4z5;89HKmJw4^uy>Bi}bM5sCn!>{M3-1q* zu9ibQU$2X0YVJ~aHA3>*Ln2$kCHC2PTpg%JlsS>{Ec=8r6ZO2&71urO*ZB<90Up3<0baNX%5~+SA6xQ(op1%X!Gz zr*vk~V&2^=BTRY0uaB#+((~z^8x6AE!X)324Q3R-mrpVuBxW1tW_5^nW-@Y1(C;j+ zpne2mI)`_Gm~;iakxJ>UH2>7K&j6eKYwl?^BKU6z@7XUReF>9<4ToRA?++w&0d4}W z>YKucX8XU;_F2MUW?*`MPwMN)8qAM%aj4fS_KkhdS6VA-a!jMSa~=LfGUoJUk`kel z*6PXj^r#ZuOb=aogoP9}-Z5X7Qo_+qlg;Js{mEk6usZ4NO}v3)UH2TKrU#UGp%(iP zoYz-U=C`dA)v$$Ymc#D|?H$)b8_3-);6|3Z_7@ApA?wDTkh(0W{4?*7;_9sKyeF;d z1QtErHGe?%aSgTC)WyT&fd;SR zU+siEFjQvO<)`%A+slVcdA!?wa}B!TRH21dHxfQUc6-MNwjUN=_mXRe$Bv*qtGMfp zP$3tsX5O4=rnvEBRBPBn9gS5g1xO#wU)Bxb8aj3*933Q$}F+GOV6DoB%*P*m(3wUc&&5tZ=6n_W$vZR5Hig@Or-065!Ww zF}61EKFQ+G`#|dN4jF1iYpOZSyokegC;t&RsBFdJWYHFVEFaK)Fd8xiUm(;R076%= zjw_9Qd&Dz1S37BHy@{Eiy|t^#7n7~q__AT03WvQwQ7Pwe#(Fs#4`x7RU-n=#yVa?B z`ykeyH}qNMs2m@@Cf^;>(cNy?nTbjoT+^9pD6xg{(}F;A@@4itl^owtrZq3wEeD-!b_=)2FB6P6q@E8M1CE>!nV}7_QGRE0)nWH!cafqLhn4T z;+k-HlCqM=)mO(j-xNiKJoa(Qh5(MPS{?H!Qn!?^vsc*Ob5*peLnJFP=fYm#?5E~j zd(mArhTAU4tGaZeIu?f(f-4!Q7h%JVfiKzlL}ew7)X@88w|bA4?L$nIy9ur;!O4i0 z9AP#V6N!eF2>g!!tuIl0Opkv@$Ertx+1XX_D`i5v=DrAj6c-vP*F8!YFZE^)Q4}Kg zFa!>&4dVQLZFfEpQc`tLeDFl;L25t=$&J*CcA3%xE_sjx4s-U!|2i$u)r${PxGCB? zVNtPeT03Q|^cA7#9A@1I03I?5=Uy>kBk7|f5dxGh1GT|eN%{Cce+ny=8*Y{q{i{19 z;l*wmpjzM%-4Ia(Q;J%1psrj&F-%g7a1`eYq*CB z|JF=pv9=3Jc0PSgx7gcFnVuP=vXMyFr9M5SNua*%sa5GtrRp`0=yM_79M)WX zkvhmK!#8|^VC)W@za)`feEW1|i$O1Z{`enOT+`a}k-ZxuM{>?QbG_;$e*u=Zhx3h( zrd&D&d7r^^AH6T`bZ(_WM%4x*?NM!|V(d=7~8sFO1VB|O|8=&-A6D*^!#ymtR z?o_1W=V?JP!?otjNUm|Xko+Zs4U3EqF&3%_B!6_Ei4hP(zo%OY^7q)t2d*iiFuEHr zy3+^}4@UsZiFA?DnI8H_#XUr`;m7arzSKyJhiGDV^>JS~I?w@%-2yob2pslHD^PSd zR0K795NK#Z{1=d*PkK9By$94TDWJH^eB zRf#`@1wB$asZ`cJ?w8xVM?_meHL|=f2kaazy8fd|#=;*0dp$ z(hl~h&dDJCLh@i~r~=VjuW+doNy0baS^+cwZ-7;7?rD z{d}TRR2GC;-&O8ifsr#Rs9D{Nh%uDBK-)IeoB(>=acip5B-x(t7zwB*ET}1xdP>Af z439VoilzmmCX?b{!Y7!lFOX#-gEe5^Y4{u|HFQdrChU1T z`RHD>oo`=C5>_jb3@`ki&?Rgd)T`a?X{+Z%RVJsLT}&Ukv&MKtr>A5g=FWxF zY|0dM``A==IXm7z+PevGxwnBSZS=0A3n zEvTaRdX^3H?K6C(J`2+GCOh3a=%@QLX|RyYwkpk6gXt~{qXOpkWLv^KDUt`=3OOPl zSRqacA(oft9~4V>0lXuhJW>fqPLfk@AS@TzGS8sip)Z<)=&7d^QE1`($9!VfA?ZQ< z&Wojmq+8PsE#84KFFHA>EXqPk3wn^WTK0R9MWZpoYVAZqV#@rw9ffQo+GXaO)~Ym@ zUTrpK>a}InYrd?Mi3X6Ot)nA~VuyCv$iG7d^Ytey%!c&Dw7lw*_#UHq)YpamNLSX> z5nFO?&frVR*h{+DTU{d#L2CX!Ue{pZu_&@OSot$fxjg@5RqJ54XS$+qagB&UR_B{g z5oFd&0vTZiD%!l7g1{p*%Fkvc;(@(AiFnJ!&NL@&trImjR8Cbm8+Aw8~$P0l7? z33CQ>Sq$9WvpM|ThH>p<#dz1J{mKigHnr}v&v)&eMELfN7>v?RS0s0H#-{4%qmO5k z**TueMwQdCwpRG2aNvFPPg<;PJsutx8=OY#F%@WW=(V7v1!PhN-FGG@E63%F9~;qO zSR9f8Y!g+c#?5xzo#BmeHBOzC3m)m&a6Ewc*2VqDc zYJ^}=o^Nq%q%9Az0gkIWkIj8ja-J~X?b9-+@j71ZYy$BbWXKRBtxFze2LjNxe2Iaq zL!wykkMCg(2VYU3*^loZJxy4P z|Ge{}F6ecY0bCE~X2(7w*Rmho`_(R=>w&`{KN!&BGPhyPdDHJS*o^@p$qwJ zQIyQuF}go`_|1#29c9E7upMUX167kC9=L1HZ40U+jbks7SVEF(HM+@;4?~q7W7{x6 z{je$h?(2OCc*RzvGk*64#C!{;oj6E@MNZbyl8+GVnz;+LroLtJWXRmT*K^d#cNJ`; z0G!#0(GY{Zu;&?zxTZYUE(n3X!Xgt{Knar493%nFlidDV% zBLSFSN6a9_G(l+dUA9NafZl@a=4tG)mg{n-3UD4fIKt{js2T-(ChDv-o2fAf?u!*c zoB9$&ts~ifA73g4JN${6jOi6hy31RzOLEDi+pv)ProyA(r5`~FY(w6OV+|=u%DaHM zl5AZ5Yak^E<0bLb1Wg>q|m&T^^&`K#P~xg@yfse@q>^PwUZ zG#k8m7Zm@?Gi#`YNyD64#=lj=Y6;`Jt}lF7~N)5IlSjLF>0+I5G0U&H`4r0WDQ zireH{m4!CPX>b!OO`Z8;-_+yHDGn7WquG;^%1*jL`v2XwBZT^ZMi^`}u(Xs}W$-I< ze51Pp$B6Na%x6x02jNcWQe4oLISQ8JYu?vJU{uB)OGwAxe~zP;ItLC>-(aXn zYw^%U&=!pSy_pShp_pya^KhCAegm+jLcr!e^p(gLH+p84u;J&~^#hQ_2w-xe$GUA) z<_1B2mJ+t6gatiQicGeIX1z<<($`^`1LL#AfsqD0kQv{(6qyjf`bii*YX@#=j*tDJ zIkK*%%oawUFniJ5-kLih!X9m6V41~8R^?i#F?Zj-wkzMkqr$!sa2lVUSndU3IpyD^ z3q*n^Y|MWAX2Mc?q#LbHx`Qw|_<7uJr4R8)at{h)^{0#+J>7Wr65qR{xTUxNiWT%Z zqfl_iVlo^DYGKAr>jEBUE-Ox8HR1db(IAAxYjDlkv5k(8i!Q?e<4rfk=G?~LjD|pl zboj76)EXdo1cre#(&KwhF~s&u*B&LF9m+SDUlFDbyFN`L-Q#_7%}2n@vJ4ZO&sg03 zHSgu`<^VpoZn4RI*;~}Uu63m0u@s*dH1fQiQUOko zZZ?>%hIcyVWGiH*E_o3T?^qRo<3;^PlE99}%KD1TeygniME|EX|3oTyHUNMcA$AH3 zwhJtsLz3{O?T?1>`*B6rew#(|aFO&YXI@M_PIZ1<{hPwQrtAg=tTrs*aTCudT#99@ z+SL_hLXg^BNIrYWTz!8C#2zw)*h2y$+Zyp2xHgq4swpG(eE+%p6Rm;ZDu#4K>*6xt zUtEL=Ox)^%%uNArGrrPP`AD)Y6nCJd&#U4(c`T!-<NavZhApp`cSjJ4y+k@C5C`J;#~FjQMn zlG_nDJ#l!SxM};tZ6LBy5({>qD_dDc)Mz>#v*TX>>(L24G0^ap-x|KJ%4_&hO8ZBZ zR#`tSoHsXzEWH%^h)3*3)2*JWyjMsjAu&v)GfL?qvm$9(sP{#Q<8v#?KnQ7oI z8yy_9W@co)S0lm*!ujEYGxFU~et^d(`jn|WOH>IG+6)GTXvl-19`tt(o`1>v#?b%# z4Zu8VHnmWa4WXY0Q;n5U8C7!}It|@;+nv8A8$DGGMd^7wZ*Ay=XF})50+U8{0$)m> zTdzo7!TXBr|M-=zE06{b{LbH1z=4_Txop*DUsdn6{3VC*Db)u1JD*Cl z9*Bla9(RU&!f~Zc^jxPRIph);o$9TrV)gg_>p#!7-Nw18Q&Z^^Py&OGkuL+sxMt32 zULewE^ifQfbL6Diac}NU=&AxdhTDJaZ6D}Oj}UfZjnU6dUjQ8|!{T5@pVKCA?1rY3 zJCD$t-v0wIH1~Nb{L#Da*V?yb>Y=^r0UrzW{Wln@F5A$!S1otMACe=zRNmJQwU;g7 z=CjJah7FWDgA4{AF8`v$O4tlyGC6GV?>vCX{SnXfWu}MNb0c zuy^k}u$|RwstDyGB#|iHt|iPobb?uaQ09lnEmnub#~o4q_UmUIR~q@<01utGE3!uA zIIiTOsm1eeadwW2HMmLE`$p`yK5sxfG4!4L12X__SZ#1NQnKo`1TudKowc;Mn(_R_ z^#lc{1E0(J$|yH-kAb$c4*oF_1gADniqM+b*8z;Tpn~fqkKan#28K!aPmMiT+No}E z$Ria2+q3d0VH@HD?24EJ_xY}361!e8Sm&EI`HLB8_VmPu%Tx#82aeDFhgICeC;It$ zf&IHCcVqfEbvIzsT@`cirwmopHQ}aulp#`^@-kbU5(sC2l=3_wD2o1G^Jo~e8Mk6{DtqE_DO&BDG6IxCEu*Gsa}M@C&c&w| zp6#pFDV~3x2h|%q-Bw+nAp07vi%z`A>iv(hn~7gsUGn-<@3^LmvrtoPiV6Ly7%yX5g_ zD6CmE+KF%0Q~=xFiiiMme2uZM3WUCMPi)ZNf!1S~?BT=6_XFQcu`m?eTy6lwb8C-z z%6M+tn1|^*g!OA-VOg24FtOgUZMp!_i;yQxexB-k;{&KIden;`Bl|bBLYf+?r^$)@ z%o-#Gqe=IxX81HW&mC%&^+NK&K=k+zorR4LezY%r$04I>h!`3k!T!)!bNjG%%3GrZ z2y>m(4Sx(MO{+vzpVj*pMi_3K8KA&5t|hS-E`zQlpspx8cKo-l1eO_f8dQj1@e*{` zRv9t76DF$*UW$nY1RU}wCKgTj(Sj^>NZR1O7zRguh>Aj;X@%WdXmvxBzD$0{eOp zzHbz7B01sjmYk%4z{XEA$9AE)@Us+3-A-z*b400EP%ZbWi!UDZ76SEl<%}(_H-M0> zq=JVlS0zj&6bHOt%;O~r6DO_95>5^`7_j{2?zyz7f&%XT5C?f)C2}jIG>Fulu=NiX zUNc$V2O&CUu@Cgl&g{V6MuOp@8Fzr2(v6rW8|Ja!EXYYDlDnu(z}J!41B9J%$Mk0) zCz8YjCEuOxPE2PA#gvvTJR~lyT!gXAs4_)T=0hVxr0oQ)D)82@D?)M`RI3n}I zqt~2o!xo%2Yr`fqYQT?CTsz4LVwrJ{sw05NvZhiS0T=PX;z^2H%!1VCEi^~=&K(zTb!NH@WXJT)x7_F@PzvYA zuW7hCQ{$lYW}jmM6xO;yDJ&jnhKiWGht$No?RhUG7eGU|N_e(16TXcH!RhKL4bV|8D9u$=aX;VhF*4<7VDLakXQ5~&>o<7Vf#>k_q zCcCmG76R84&d+^261KhsO~^sxS>?17YYstGKFkw>*RHUIVD*7WHYg|IRQuSzE}JdZ z(6nESW&5V7!HZQYRXSg`z(y4M099+`FRTP*zdvpha8nn9H( zD80dOiPMVuEya_>ETm$|*WsSx$q#Tx6ttCYf3FH)Z2fR8{f1rQ8skXBrLYx=okqs& z*wxW$rF3F@dQ$}`AhE_5JCCdt+Q-sow%^8DgWoI+ZRc#Yx}Qg4^Tmx_TRwE+#PxSv z1Pd{5;jRH#j(B;-9})GQFw**rSebGeh} zpLx&rtmZI|=mCNNzT6Hx?EHZmZi3HDb6r*Zg|X%i3ub5}qdLqJr^}qXHQ9xa4@hIM$GwA{ z$g!9y#oL0ChCpTZ#Mxvs(pM6_0rI7R_yp}SJ64KO`nG$1-Q9$~8%2yfSwAkQYreH3 z#@>Xo%Ps==dCA?F_<-~vvBqeJNKF| zU9TmxW!_@vF^yMrFpX^0Ia_mKrCLnep@!Q0E`s0Y_m9Tx0H?n{B|kh&($MW4AQw4_ zp_5`ce6u;4g z#ZVE^zq_yB;WmoENs|rnqk-(Bd$ZXlAJh0cVNRd9b6yjhf2-jru(o9DURx?g7Em}5 z8#~I@GaCifUXzZU5+)5(tc0~Nou0*Cg@neXggXPeP~F+9=N^kNdm{{X<6m;*q_tj+ z?3FfWcDdpuFEmKc5Wp6a4@|f#YawXlu2y!h`%~?z`!Di?GPJ94nXYxB!dn56P&}jI z1)dd?z36!oa*E1Et=5#ktBOC(ZIF3AJ4*{zhcC<=r8)rb>jckK`gF3rY|?TJQFp3s z`;)f-NfMcZ(RA)vR&TUuPIg5uG46zKR+`D|OV_K_NefIN*<*Xq@{Of9gKnb_fD7Cn zP`X&2+-DsIlZ8ECce_=3@hQ2`FKw2Cfqirc};?mu@15CiFJMhFCxLAN#3p@Z>Wm9O5Z4IwD zy0;y=s&5W2BC~FVM&7eZFUjq?dj@`fq}2J{wUYotS);M=gZVEFMs>pQ@RE7LT-(QEw;SM+O zqg+Y9biULPK}o1|=GC5}0f-x$b@pL{|2~)VrO$j+7;o88!6=rb6J9=DNS6;m_c7D1 z+=2O1Y13Um-c&fA5XM!g;1PX!_%4Tt*D8VfL%c#562EryANL`%jcaY6-bvO}Ng2L} z5v&HTNGP8r>$z+6q!?|Yr^}(5cmYtFcAAM2x8fa|3wO=Upo3-x~ zl?HDO*tzZc!K#Dm#Z2hDVinq>1`tE?0Uv*#Lz*e-$er7H%EoNfdrmJ{7VZ(6J{Qck7Ac3LW z14g;~bz26N@|2|>M;16e8>w-2gu$O<^E_YCJ127q1kFEm3 z1{~(F(b@W6nN8W7#zn$>DF+2_UQqvlct>!&$ob1l$>Hu%y+qT{2Gxw|-+6+49cb;uFL4 z%NiC9vsYjt&-uqf3<4DcH>$rl$uiivz39=&C#3tQ1((R}0~yNb9w37dfQ${vP+X=N zyXONiyYzFv>|7hQlydgBnxn_x!`Bts$vPdI(h=1iucR}?r#1zUFZ<;OOl|PXXIEpb zTjw5JLkn8g|w!6@YeR({J6(6#JyKi`884O zi|3S}5&|c)cy|(Re(f)k$17+Fi=scjBQYG`zRY_FWeAos*Id+@%Gve@|7D#Di5_1e|xx5fEF=G{Zb zmQdg;36AdLt#woO$l!$g55R4YJ!_mLYMk>L16ab=lxL868y}_q=6vCzorT5}U@pis z+f0HQm2bwO7eSf{x==k4eGUP?J`F83E41FdF;OwERwS82_R(Brx zSF!&HL)_42>NWI22Hnx)5h9S**AaA<_IPSN$aho`qS4TnIc5Xcu1WS#Z`_+{(dlJ2 zwx>7nRK1Vy1|yz6PMM!17y}alB30(OK&3^>y^@3Ny)B$&gsYbB!;334o2F@ za~UbLR1my?<9+6dlivb(>S#ZAocpbeF`Phe7H zD+>#>7yUPG|MRYcx|5QqJw1EHt+UCt4-g~c@t9~x$t)EKk+6AX~7)-%| z5jxKymQSpX$j47;M%-7x17@`S?w}cD9VtigW&6HY)I5wW-c#Z3F_keuO5@z%GZ{r2{uDXx7u_sY$fr_9h{JhoK8k1b@>JWphE|_ z3&&hs=vX}y^9qL=v6A$Arw}!a+Xn#@A$I-`Oyz375p4g0K6_|2Hzf$r$^8b(XMONY z?*;iY$EI2%&>Ly% zIuft-`4+JP(I~=`G(MxVnv0j5OY{-)%PVOVJ{-k`QfcbOf|7R4`AnSINZqbcqKLc1RHzZ7ZHdsn1i0f z8Ke;*lWDSh?_!W7h=5>nh84?y)V3?`>~Q+Hn>4FfSGrhNR~6^=0nr8vas0K=`y3u! z6g`f8!!F{YU+y&q_N3hjz}0`Z-!lW)#Pg8iC>kOURpih;C-v6m-4zlZ!_Pal1~ac1 zvvKCf=03#8>Pz&Ugd2_~j2hO8Tf0ge3=_Htf8f@@LUkxM2x*O%PadW10*q?}!h?@Y znOK4P^H6i$%|jhAlI3VfvNO~~%()=c7jSVAik0;t^Yu7m!5gG}i6e#}>h#-+pJQ<^ zol`I!mDY5G@cDOyz&+E>dS)T#c~Rn`^{lz}K8b_YJT08xGIe-FM}^TRBAtW2J3aOK zus5Bg-ad~ACTA;m_Bc!(X5`-HRMHe)PfEdB4lmygFUqlK%}pqEi_uL>UFc+r`a=)W zgAyi!5{7mpZy(Z@&7d+-QQ&^x2u|Cbl-QLka}~Fy(Vn8e7IBE;_06o;dc~Zc%3j-8 z<{z-R$f@Eq#b6L8Cbsx%ysB|sJn?79%?x>3(FBUKqVDWbZ+U-5R76Nz!ndfce{Xxb z&$U1~fjv9#))TS4^DKHMnOjxf7{lp~iRtnT#2Pd|R)OsSVc#XeSxdAUQ$9A7CJlog z5^KszKCmggpxY^YO5(xnzLt}l&RL}$3=@QNy51vqbAq@9cAPv6JzJ`@b?TTar{*%7 zrq_I6+W7~{gw{pfBO%~)U|Fu|i;HTsTimVJvw}&hPJEUM6Jw}F#l@<0Wl=h4oNo*WtO?x~Wq4e&(&@T`~Q#`FX6z+<~VZ*M>l}IW~&+ zg7f~iyw4cs*5DCuRv1*{Ku8I;23U&x1xjkOQ6E!2!UV2zP6CQWwz9+@B5 zF44r$>zn#@(I>It(^D0Y{~I_ZUp`6C40vMzg>~IQ-(16MdWxuhhXmIL%x`hpyrO!~ zFUhbT#uv9P?cy-?|B&Q2!)YQ5TJ~sHDx=OLSmIzoEp@ zExh_CMtCqWT6s0QHyM?V1B&Rs;a+n^+S z(B@3^%c**&aLh?u>zcU3VXG_Z?tG~cY1sXaIhhQ-!!mImdeVmV70!g`je>`P;Ng^= zSHZNyHAaA8`fuxNVGh5B@i{$tPPNVow&n?3=ty$7rBW$|Bq*-4b44~^n0kiZcu&$=> z$wN$HpmjYAA6-<|yCbyi_MGG~m=)dmRQtId+b)Lvxn9&{#7cs|du4fW$Ur{wK|U~l z$@}9JcZ?^`IqK!nh?-OEU|>L5vK4A4nR6ahNn0-Rw~nGTyDFppU#T5z(nVz*Y*#kS z=%Zp-+R4Kz4;eP})^SX#Fm53qt1})p??;v`V2=#8tG*g z1)x~t!YY?vId5z+3U6;-m$d2bNIezS31ViyzO3Z+_J%jl zDv4PWty`M;R~jia2lv6KHav=Yv_4B{E6c}8LjG_ybU-g|PSXRiju*Rdm2KL+xtr*P zii%wt4#FWOti3yyq1?pU+Ooe~q{1Ytto-cuhom?|XJW5JYck!l4BL#F^zmy;{HzRL z{Bq6Uv4%KQq$711#qDh+MUIOzUt>bAgQspZirEX~=o4PXYuTaJY52GSF)rtJRvpip zt@LfiPW|z(&h=uVsAl!$!doZJ3hO$tsX`0@*IYjH$vHkeHudL<{&{u1)8;>)M<~?0 zkI!nFH!M$|4~cAYAJC62YN`=I{rMiQ)pGflocQJ51zv7W>C?ITJ;XciY`~x#=}Fg< zpYDmBg8v#!MP!uE$$DMkt#eF?g|}ll0;rGwTyrRrWZGud$E$&#pkk-ch`0=jd~}`n zxtILxQ`z^MtNp;?z;<;vbZNX2huwp2<`Qt-|1?1N#_h7~{Bcv7`_>5Y?`5iZG-sKh zC9$`;oC&(~g0onE`jEn!7am)l5f(;a`1E?en;}=RJHFgiGdtih$kz zvK)mdHwPt3(n4=SQT-Wxxl=btuS#tI2VC?}cVHqKBDcG6`~bF#mcV@(oyBH@@IF4D zg4S^QkwBc=7b|ynaD=vHJ&===!vF5E=w;BJfdCcGOi0y*WtqnlJf+721Eso1Gn39z zf-6i>@{0HqP$)8VC3n?o@C=QogG;aS_IP>E518pOIe$>LusMfg`6(^foZzz(9G)5d zk|HCaXAJ!5H zNxLP$Yr-o-nQP|txj8<=imOMY{=LIV8QI#`c+X1?P}UnRtXmClMsyV`Mm`}+NzOAD z2n~0kntF*@5?CZ+pJ`S+>9`kU=rRul$kF zQ!J#)LYfo5-(;86qVbY5KlD!C$Hzanu!lUIum2kKLnzpMwo_APW=xuUzz6{uQ~wA7 z;$H^uu2!K|9uAX`RV`Rd>|~AXzToQg9{t&}utr4g^EBm(UD$d?gB}u1uHva2@3D z7u-GoJQCOg^>f2cwdpE;a7DtJ)XmE$*0?=t`vj;9=%xzK zZ`*Y`)h@=B&;&eIEG=}acoo2%0JvNSs0Fz->)XS9IC3VP2{TTb7m4Ai&A0gRM%FS!@kr|fO!M8?6IcHE??dc^swS4G zj5FCA_av%pIcRsOU;&`xh%vw%=>0|F8-624z%Z|nw$j73gEDW6KbSr+Sv80~5yrre zpm$CEeHlxE7dx&pY7RWL4$X3V9EIw;6f*XagEzMX_8q-vm*v*(6+# z)n3D;fG*nTTW1_sPTC7C2u^dop(>e?rI%##&;Z9!k8H)X@c$F-dA|M%@3YoLlN$KP z+R3P^-FJ>(z;N7Z>YR=H!hpm0AHxd#qMYarWT$m5(rYi9{XQTAIC)+!;kg(n6ZB>@ z!;R}=S|>37s{(=gcm5jrcr$o5BE6?DGWOR0!P8B;A+ErZj70HdEZx2RZ!7;*UqW3t zfAL?B7J|NTaKPRNaX)1Zo#}sP?ZN=XMqaWWoA+-lqV`M80O~MO{mGRtUV&5B$U#G? z5i*1eY_vk62@=BUu8*A!`03ZOP}lt0f7tR*2ls_uHOFGJ>&__yRHFF5;!J>+^7r1k zpBy&28!+|cjXgo4u!o7Kei}6h$mnnFaeoPeQ0}FI7UO5;s~uRUK~eet`ni;Q?5aji zWY}<(w#Od|=0B}H?ZrQE({e1haH+%@MBa6G)4&1q7#=3>1)y4gOl@Q4#AkeV-tyA~ z0#Ku08mmaH6;UX65fL(|zv-Z(K-4U(XKPNFM_n;z^+!k%|<$O3eW`7ji4K87nX9)$}J;(UY zxrw!JIYev0R=MR6*n&sh&%%9M{+&t{5#L|h?&Y`$6%2LIJE4Ab))?`fz$h^-US@eb z3)&eJ54<_|jyAsREb;e!HI{;qtUJ&5Vu{_5yA54HctXxWkNgg2vR@Jcdd~8^sTG68 zeBXS^-Wxc-Hdt3flWtdU%)nFeQ{;D-`tb9~Lrb+w3VXree}Z#r$P$k6_8kj%mvdOK zHKp1al>Gi6&DfJn-!lVWWj@$ZShzxH9g}pSjWwQo5dP5> zSz+F=V$eXLBR83+P&j=hc2-Iv?!H87w%xW=h|6=GiZ>%A6Yh}&rG29doL6_}`Pv_~ zJwfNa!yL-;#4m4VU3oq7Cy4`#(52*N71K?U-6vrHs*?ez&=gXdJRMhzH6cqT8rWjN zMx>2R?nxiqI?whIBukv+rw$_f_NDE6ah3V=(`pDn7mvh2q4JWKUh2N|`vA=8vhz?8 zjpArX2Ssvq`^1Au3}4-+o z8qBEzn22#p59$k3Y%?2#rp#bx(v{hwCEhFY-3q(%D(|-TL9H<5`aBWo9fDTneO|3- zX&ioy%75C-xNVe~i>?_v^LoS^fvi@V$(ULO=uHBqnQF{imx~jgS~I-?r<0~x_Szt+ z=&qtAO7FXK?@)tFEU#X3+1|Iljt++uNk@Y}=%silo^8IDnG;F9TMuE5ftX|_J4>Qu z;PJKZNi)TPFVoX%21Y4)*MgE3L6-*~IPC~W<)vA%GNCTAJa%BJqwBYuDk&rR2=TXA z>ytZ9OnHC;%>vmi$*mG(lNR+7CQLTWPj=){I+94TpmRn>vWJQliOF5_Y$kS zKG~U*;lFKU2D?S*JDMYR{1NWMqFL9v4f4a(a%8~>vPdy~Y z@Ijarr*S!R`E=^78|ig8C*X)2owU!;u`Q1W*U|jtzC$)+TB+^c9|TD`dP6MQ5Pu6Q z%Kmg~){(O4=F$1c4LJ#X)jgLTziDMO$ECO<@tE)QsBhdZ`VrLi{nJTVM+u{Qd1+P+ z_^RutWCQ;zWB1#mDj1Qwdq0PNo{wSLgR@h>yKNwf%YU95Cd{6BaUD@$o~ zY#!VCU%58xkY;o*cp+D_fF0v>JdO3gv$$;xoJWRhJN;(YDkt8*$~Adb-Q_OdjA+W{5nACHfrZ~-mO==6!!cr zGW#{8Vft+`8VIraEzDhr_1x3?i5nlKL;oU*o;{|1bC2)625u5PJ;sP!e-E_QvDxV7 z9a=_AL)f;8cE5bh{9ivZsUO1j$Lq?mEt$)1}eWZR1i!Oq; zCbH{cI}=lL!j*yz8#VI&_-!3b0B>#jTnO6^(dDRjNB_J!e|OyBsX`WS)U#yUGMNbOuYo9D46nrm%TQCs)`#o?=1Ba@-%w27eC4fGD2!tT`iCDt`D1jgVCpWtOz_TV5FoEHT zsPy#*oaLT?2Vy}whyH-rJK)2m>}6Y@61UU&5P?t+jYj191FYV9;3qvCfC$j%MKl1= zD+d62_dq$2{*z*-1NB>YqIj>RgDqzKq%<0gA$5jx{Rd(XL6dI4_GPpxMSB_^30V3eXe!VvN=g%hZ1wInf@rR0=XK7I}X=4W5^erF?C9(0!y#FZxsbgC1Su z<}P8o1Tc{KURroRE{wc##gX!i0T*W|wIkUmvhZ=H?L`2mJas^&KLA4R3+ZoIdw%8I zJAumahyBjl{g1(K`p$=xvOppqS+7z7b)5i){A>%j6Q<)i2=|q`YgeN|NASA<-A5sJ z5h+JU@cRt)(45@S%Cio~R|<-S8{FZZw;PRo;vmF0dj0_VdkBAt?kWI06xMaTKSRWU z3w9R^JVUmK>?7LN{X9~3fIJ1k{n4%M-2*9-KWEf>?m@gp0hu6`dSpoW=b10|KZn>M zv_Ht1xw%DXxemHa6wnVRZF}H?O^s9JJ*lt#57z&p@BCc9;QG{>&MBMFp?D#pf`b87 z!u3y&@VxxqRJIL*_M_X~fNy0=KhLWb_;2;)g=(@h40I=)wDo}t{Fe&w%cy;nxqn~7 zmA}vjZ882^3-FC6nxEDG%P3F%gCx?})=&;@rql}@-@bpypuZ$#vvzi9n7#_ zo|)6hBqT?ml>A>0GYDw5up5WG|KBb}{N#?=fnAkhDU+iY{^FHBIq?;N$ryH<_XuJery`j?HFj?_6rBg8*bZG*_;TSvXQZ4@PtMG3?vU%p3p_LKg>Uf zhYxw_3@l`e;I7I<%VC5%ME@M`uMoF+P&5QCv9UA zcCrXv_Gn?Y>W>&IJIlqq6WX~~uAueaZPxYmAIFgQ0NQ9Pb93ChlyLfx`{Yz-8XlBY zJ*qx6jP?S&7Awx2Q&Xi-GK9;76*_&#H91BSv3bs~G`WmlR(>dMA_X88`d;h(neasJc`XbpF} z&OKO;BA<=8-86x?7?+1kH)~qymj6!LFi zTx$1WsVNrL;(;yvDU*t=Xk-O&SSU@trREj6;FL{I9QKdgrsZW6D0sU(qA&1)SbQ0E zEMf{e6_aJV7togvG&I^gPtvvVJKx^Q6!9#}FjugU1i&R(Zo{Y4(Pyd;@pQ`$UAkM}j+) zOx*iS{kE|CC^xMzDJg)9>qNMVjhMG061F>lZK8><}rjxO!prQZ9bq_aBqvi|A?C;af!Nf<`^qjEuj} z`3+FCeM`$d!K%^eIo9nVAQPA-nJ!+|ukOu7o`iM*=&_ds0G#E=f0?b?fk{X*(V}0C zKJjzP^GEIj%R@s)uMs%=EZIeac$xVA1 zGJAM}^=NDN!h+>y(9zr8I#sL=?w)xH9k(+>;4ssvisw#DMRF}Ppc09~;DXP@_Fr`r zS9_nI8zRwvtD&AsYuX=1Cf#O}l#N}!d)}?;s#IrQrY;LX=Z$<};{bk*>BXaKN|O8i zj@K}B2te`*QS|>phCM$!<>rb6b#di33vhx_Uh3+BWwt+frSd&hlR`6L;W^fr^Iv;w zH!*V_OX~&~#?~!O?EpL-ohfP2Dzc;a)3wzAme#42dj{M~(qodZn7;q~5U%Yi5-k@`&H3#|M|LPZ?etYQO9kex!4sm8q z#8<}6^nU+?7XRSvA7_XPBM9I-1e}I-e|)?0{KGYj@h$&>2sJ;Ti(WvKUHf;xbR8=~ zlRE~wp&6)^JBV&@Z9^ulc|C zsci=>xBcvbFDV}@AukI6;~90#h@eF5b)hdoDty~6TqOPJZUB8_x>}w&q{Xc@_!|1M z10}oj!Q<(}GlkGMrWX#{m?W{pDg!9TaDdzOP?R1YCTT~ZzcMBD*XQ$~i0+PRif#J0 zW`HaNt`^TcIJ;^;gY^4eHM2Y@aIIu_J&&?uW<|NtSL@ncN^moYz&*N`qrPks;jF); zKI|6X#b#+W5q(0l?o0_Um(lUi16F3k(XI9A$v%$&`gdo8@qK zWb2NIZ1r!vhqZg0F-fUI_6z;yFl=v8CBd^Li9n6@jL`K?B9&ctcy-9U>nWV=%b$_clZa=R3l$%43Rx-2@ZZ=6i1R5nS7S>w*PS!RV8UYaOvkkv2`gwqnSBV&IjXE( zsLt`LdCPWZ4Cz?e4!@RB{!`KXblU$IklgwEWV*h6jJ<^t$_;c6Pwh`{dPS3OcYBMpdvX51B8EKqTAAntt}n4st#_`T z70kO7E6+K@z3bcNjd;BEuo3=as?#;&7?Z}x>#+__T+^Pz7kzbk{;)8+lrg?|L#uyc zwU5t@Hi76G`4ix_HgQ?n75GOHI;HWqQ}T!-IZdw`3E32PqZuTjy*MG2@}%sQ=4Z<} zEy^|~^|-l;ylHYyR8mGZ)-sN2P(J9KH|A*wYnnH@DUtiRn`{opVcDcUPzUn{E-=Z) zd4JOJ(udf&)){kARL0w5>SZNvCO?Bn5$g%z_D6fAdn0+amTy}Tv%fBroG6l! z#VRZ{`*Oy5R*A2g_hIFCZQ;zoy?90Io!Od-d6Xm!xfmTRC(iz{UuU&?q9!TF9YeHq z?S@N)WLB~>#pMr7@`d=99HZ_YxuHnvRKP^fpHk}Yq8d2`DqwO+19BlgG;KTB`~Px^ zh7-c*1=tPZ@Pk`7@S8VIYKY;tiKb1q4mC|W3$Q%V;!T^V3(nEJ$|koeN&?;33WMb0 zTU;jO*(~FR{HExIX?I22>Ps@>S6@wYc!iRewwp#i^{&2s;_-6D66Niz2dDtwoAn95GWYl@^C(#q2$wFPs_ zV=&fk?`iR*?#V$Dnz6O!I8R3tIHL%^N$LMBB+xxn;`|o0h$4+uf;bsatBSr9ZG$N$JDwsK`|elCG`zUNzTO?+fo ztC}ykWWeZkN`E#u`!RQS5<54PzpT%k6U_HnLiAaK{-FXvytCVar!m%`lW>t#pl(Pu zIMaxY_wjN~PD->VVr6esQ!SoRs*|Km{sXHTWpODS)>yVEl-K$+Hd$5Ol9x57+?wz1 z{9%=%8H}U0P`rMvfifDLpdr{WK(7t^2dUk0}x_g<&VnyCd=LHs%N=jx^YiR>)75`?c zv6vc}gved=nn%Em7i1NeKAP*5wo>;B!`SamJC!8xvncR)PDPjU;5ZUY`tPARrsa#C ztw|?+IpCk#%zB^9Lo-g)i!Qb`K9}6|k#&a@ zIf+Nq;y%Xc*^4})>lk*08qX~g)RM%rZH$6W^r>-bRW){a(H~pZnIvnUo=04~%qpNv z5a&5M=7Sa^>EC22l3a9hujV&HGHM;9$KIKu~-Y+b9K$PdexB3mTz7-MHC2bNr zD=*yU>fp?Tn$aCslAPq&XF1jK_DA(8psaAO4i-M|Y+X%relYE!HSa$*@n&dkj2Gf| zb2lnJ_8Ca<-qn84$n_&-+saDozRLoyp1`^;uySRwGV4g&Xe(DNU5L@6X0j*bZf3JP z5l|z7Mpm;b)0Ws1M21RzaCpR#nMW;=u$Bv1D7JZbxUZ*xiwV(lASKkN(uLQBoKo@!bbe~haAiM9hlxUvQQT=*lty8UT_`9g-qUp%Wl`7E` zFFAtvnscTu$F1_>SR;auL8{tf22=gw0`}S7(|#wTvOKgX8=+{H{IY#w@a`8pxNZvW zD$t(X)Yv~Z=UPqnYRn>ZH^9D4b|lFrHTF+>d$IUuRW)m=xt3K-R~7bBWE(LJ%}+G* z6c?{4YDTE^2@i#b`E0sevx?5m%8{2QmrlA(Eb3rm+8f+;@Wd?HX}3wYQDQ|_1aQ@S zVkIO^(%bH?(aYjby25xG6pIa0iw*>SdwXvDgS>3JlL}hg)*Fpbf`J-i)_+zGbR7}- z*kn6*r(8HYq3X;9)&rNxvaWIotSLT-&TT7mH5K*S@WTOFIRo9dB5pkuh?t-r#a7-R z(_&DpgaiynUo??Ur#&sPatwC2d>p%(P1|_-8(*Gw8-V5Rz1|A@3#-WwewE*9q~byol))F z>(<}ce_DMZLCD2>g1e0AnLWAU#WdL?Uey!1UlqGbJ`3{EAnDXQDr^!R_xfc z97fL8tK^~C`!r%H)t2uQ`p2hC;OS?v#ntkwoZXg`O^fIYGg2GOjqhk7bl)kVvDg}a zcmZ2*cYn@n5dUR+oKXy#qC-N6*b}BF^NA6%j_E{#d(Ks_71e|R0k>5dO3w$;vTCDU z+Kq%9Bu46cX>^&j4qBXiGMb{#!9+Y3K}8cr9QWbLj?b!IT!F=i9Kmq44Il=tk}F}h z)${p~Rd^=TR6Y2x)qm{j{MGr3E@EE7NUkyaml4K45Z1`&4v{6w9X>;}x8Aq)bN7oK zliQsD;9`$A2WpsdTt|L7pC~FJnJs%HLF6e;?$RAkAr%zgWpb4);^69Sn8{%NLE@-( z)i6fBkd;q$s@omYLOc1D?qx~QS}2J;?|&GbHPv+!>f;2H&uRQPcq=+m>aB1_)1|42 zcbaD{;W2T_t#?N+J-t=oeqj$@7FYLRLD;pip;qIm?PAU4iXX|bHA%3-m&;3|+e*;V za}q3Xr&o3P<9v7*?4wupikifqJ-!u@Eg^|R0BlKLc_UM`_jnfG*$&<$P!7 zdG_GX0n8Gn&yPLcR6AF6%kB5=d&uEh@0l0^V~Nayykq3-0lN5v6EVjECxEkbx8WeS zDyXPC;txd%n`CjxE_-|Wc=yz8Q#YoSnQ3pjnuium*@rqLo#&IWReaOu2ekZ%_K*Fe z^*gfcC@a(KaS4%pc=8F%^~YH>`PYV_k9p+Bv}{L!KAn7Bl^gc?Zh8L+V&f3CX~1#g zS}fL}%A7LLi~Hk+{KZE(J0Lm9zXarDcWbLht1XZGYrS|0ylUkVCS_+JvZ>Z?Yz>h@49r5;ZzbdS zrO_9Yr`>WNsc;q_B|UpJpC1kftnK4Crb`1WWa_!^;a`oY=&n?d-EGqUko|4}&XR=K zodJ)QaC->l8cgIZsw@kX(l>8|{LL*WuDS7>4_??7iwfFbPc*?%T1Nufl}%zY#tVxN zq!?Kw*R$cJE6`1?X4`Rrb0@R9l}4Jw8k`eXIc{&|Kn~?DdO+<0FASzjJoL<-ynByt z$VK-E`eJ-|QdV*~W1Tx2Zr$lipLQyo8#x8Atx!9DewL$B71D0zB!r%pBz zoV!KP-zOIE{o3w{&+$l--t?Ib3C#-kjw-wk6J=>{N^Q>F4sGDof`Li+9)&%U`BTGz zRs~ZBThqqB2e?OO#v_GfVF^_sYbyk|pKn<%nuZ@xwd_d=Itf>+X$h0a{V)y}d6$s% zK881<(ym-CaPCV2kd6{HA&79zj(Kkh3RIe-?Yqi$PrUrACXHc2d!q{HH3>V>A+jp# zw#JA)hjEv(GlD~k)s)u!CI?pGJ3VqzW6N_@Kk~E6tg9*4;d_Me!;cFmZl7@;n4vZP zl$(+hoqPp$U!ZaH8=R^V!xO`Ky0W?D%grj=TWL#PYO<7lj(sgXqdHP9bsbRHE6FdF zIR9kN$W5c;6DeEM9&+{qzud0VGSx^E+3*K-Tr*?#+xygzuV@8X{_#mG4i)I#S6#)N zFiJYNIX7K)rWEXMuY+G#ycR=${=shh^ht>~ge?lsy#zy^pCj3aq?+ z^a{sh&o^vDYr{RC$eLSG0@=w^V2Xa4!8~1L2nOdp5Hr?TMzl zI8giRpW{ZDZaMpkZo?4RRNW&Wz}II9i7tz;@arCz^N4nuprN_mOIxFEI!iO2%)g20 zuJozs_7e*l|efDFr>5V zA%dQ2bshii+)NMkeopkD9m++qyPnaj1wHO(Py*>gzH!nDotg)=3BrHi=)pH|8|V?R zjFs<8qOKeteC=XY=&@(JsmAs*PI>NP5H%akJ{HuG&wx2J|Sv5d7wf(jF$i3W&`66!yX9ztibVoh7um zxY>Zf>Tt^VF}7YuPx-dHy6q%J#C0z&&IbHklI<1~(`X|gGkjisB`7ig5g1$ z2ysuvM?Im~nzqM$JPHvL34$&ao%~^p_1=q#v$?^P9>AM)lNC~S=!LV1%S(u~<@ng!nv?{G-_cNydiC?BSi%FeDM)AhJ zx-ll14Oi&nXL53UF2x($pI8(6J_)buKFtjP6^P=dERyQ6&$vCA-mb`GT3Y^5ceAN% z`RuE<9jYwl+w>Qf)4+4`g|fXGpuk8ylXn+xcWkJ!suvw^6rZo1u`EvU`m3wtyLP>yOo2u)~j7pFpoU*Q(h*9hxX@lm&Pi_EKYQaKtSYy8nK@awB&hS_aR zrWfoYH0#&cTnv3Qfsg5U(xU2Yv3;fV+KM#6ib~#S9NnQ|9XY%@Tzz=E8%<%K^3gc;i+jtmlDmKKj??9wyE|K)q)@y0U~W8>w_baM)X z_i)W%BJ?U@gpumXnf&qF(u{YdqX6IAIz;|4k(YQ@MXF#x@s)7>+V)RH{IYEcUXDuc zAmi$;nlWNsN#C}8@F(Dv=Sug~M}(BvQu^|!ZyqAtO-j9Izi3o0isN>}LLxzm74*=ZJ`K~Afjik&Z|zRKdNH7UTHCM!aqVA7 z9SwF3m8v$Z3V0AWERvgy_Z`42DemfCcJdzLsKfM` z)jkDc)fw3MCOrCoF|*TPW|Ju+c(X@nvr4X3joF1#7dHR~Sskb374zr~786r5<-QD? zZSs}UANzu|vpWn==T^7ySD8IU8h^+;Q*#mrke|WFua!RJaP9l7e;QKGLw!vmUCoBK zjNLvl9imFr4;2~4HbJDwEO8i7h-tv0GZu)`Dl&sbD_L@idth8iM#NKvrPD{&h0j-qW_p72^mAU+QMPVf zhEBsjWS`vKF9CdlbRc@ib#88Vyapo}BPYKuojtRBVJkvDaPa&ELVJVR_X=_iEis9b z6P*+%iJq0ANP0Helk&>({F-C_B z*5y1ZmHcjqzADgA;4t#W+Fe{?DuT#17$)RiFFu26g4-v*%c|$pk{ocbP(#ye%1nMm zjhDI#rK4En8~1kbQF4|9bzjGCZPdLf~ivR#jl^RrQXs@ zEo|mqIPcXe6R~S!z64a>xNkEtkq4II)sQ+BaBQzXx6JpxG8e*heL#lmS#P@<89OW} zGjv)HQ446OMck<+{V+O1!MD>lwsvo4aF-G9od2Izb}6x$Ie(LJz8%8rI+5G)R*WRc zOUvTPTd#Vzj8y9E>D5!*O!eyIF5W7I1e9`-v zvHc>QFV}4rRl9=ujdF*LomiHXk<8(iDsit#pyW@|is@m}NbBkG?;x&!KKT3=X}Il3 zwbHxJ8P5UQ_NuuAUx!h`;P9K&N|LxjdS1!JTQicYPjECFtwe>H=R|?tPYp)fxz%1y zR?qRNeOmBw%D;NhuDx*Zmf2VHX+>B9lnw&^N&=(|_0Ni4`>3JARTBNLb5jG%mLPyL zk#O6iYeTuZ;;9Yn2Gi{lYe!2AGI0usSTSrBIUKV(_6b{6AHb1bv|Vh#<44~q)Dd>} z!L+XvB%iuO7^%3D-9?l=0(}!i?oJ-A%uIwwf>@FIhAGxY`KuP7VD@q>a&Pa;_V)jDbG2FJt2}g@+7jI62BJwr8~! zCs~jo^Cb65_p3>_46K)3k`83<_H8%)^e3?`iFXv(Job%TVZgihw)!DI4-z|N8SW?O zlAQfcrVMaknT5ZSk1Y=@s@|<(SWA(akLfcSM|r9_MtH7>G@HU=L>|$qvTyd3xzqzG zz*3fWn<)n!3Bs)9Edj3(?ya)w@l^bFnW>8w0&1X3tN z`SjT4k!vEfBf>(yhw5R}PPOmAI#g5LSzt5Vc4yAa4Oc%jBAa7`jaRF01?#7aa3N6Q>I|HUyq_{VFnzG?x5L0=G`E?|6? z?c2DiZ#D)`xJmVQiido@5VT_NyTp}b=-O9Bct6~4Lbji)gdEDd1qmS9d%EQ1y>A%6 zJ`bQ4#o>8LmXeqSOmC8}7fUb?ZYeGx$(xn(RJ!WfjJL^xy~&Ew>KsgLaz%2vSE&Rg z>2y;*S!Ea5d@2`l#U1zslcf)NhrC;J>(eC1z4b5zuEnZj1Ved`>u4|f;y$}FG}?x6 zpwYbA>XkZI1Nr2(n5>RcA5n%%iTB@-<_Q$oib!@T^+y9OWp!Dm%JFVoRRTAs&x$0h z(yG7E1Vb65MSJJnbXJ7xF337ff+gB^-W_dn9-p5e50WSXH(hJy7C8)eJ<)2P>MZR` zuY#e~;cg2)?K%~bcB{+IdVLkn=BaRiu0@4cy5)@XCava_ddQ#tSkoJ>92;bx-GB4D zMr(<9+Uk)q>v#MSy6@{A7hg2n;5qg2616+%YVXT7XMPO&6-6SP&c}dRRGaesk+dD2ijH$wxvA|f} z#rfqQx0G_;RCiqN_Z^F-!es^&%J(B|0u>eC!c8;M9y=w;3qD@+h2Y#;GW}EtOJAUI< zL=F*8B+L+spPggfQhfYEOe7{xg^hd?H~bK!pg>`-rtR+2jPaS}8t!9JW!mFNJaOD@ zwEs^YuW#)g`6`@vX|k+KFZsk(JT*SO5m8VjsmwLe0b8NP%$qE|H-VSfA9X+A5g!JM z?c=zOg%fftVeVIY*h?%r=g;qo-aC0tDJ<{cHlsiuvs?0E~42cbY4g^v@A zaCKYLBFaBkDr`pzOGqdbI=oFz82_I7x#8Bkh{i$FujI76%EMHf>I@7ihW6Qbs5kLD zX0$dbb|%SkrtyAI+Q{-)_?r@jf)6s>k=*An`lX?7KzZRM>XFq|SO4<{##P5{;;KYz z#YHo-jWVw)7hq5l7m$fl)Ii^-vY9}oi;&kpRntHd5olj(@gnD zI!(3pC0SEU+^5SHd61*I(bRUn?fmAv2D_3Aoi5yDWhtEa<|&~yr#Ez0C|k~?S2x$Z zdhUU@YYL+0{sukc_+(;5`{~)FvepZJbH!Dn7hGUjs7>=En^e|~^BzH!@VC>qO9wGp z6=N^a2ZV-((EYC??r>6hb+nqwH2PMyM#$5QiFwn-C_y7&lSal5p6c?EkOj5Sd%%7R zQeS%mrywE;p_NNEc-Af_U$xS;YK=Ef?*K}Ni~?1?%~T%C~;{8{H+1*lggjP|5W zm~E;qQX-$=a*ZZvzvYH~+x*Anm>f(@_be;M$+UA3kI1Y{vg)aObSnctXRA9se+e=K zAzUJw^AUV<(p|<&g}$X(0t3WY8Y{PPS_2lL{I(o@(n)~lL`!bOK!ZtWk=#L&Bmamx zi!n|u>RV#C?jpH)ab?EiO)R2~ZwJ45Y-Y>A_+5O;1E>3PlFFZ7mI@cZ@s%5+Vhn1o zKWNEw-{mk<^P<#A;Vh0tqi?dXkqDH5=+Kc1E;Z8u9g6g;6NljBtE`}<; z==Kta3#je$XP%Pmd(7AnTGExGLNiKU@$(+h8Tic{*9H@<3>LiRET>CP8}R!IP)dF8 z@SL-G37Gh~#S!#LTFa`%W+rvuDL!?Zx@qL42_xoKQNiU?W+i%<$q1JIi4pNRY|{;( zRuTq-hrh+PUjsfs+44<`SuNiM49Ucu#I@+v=z<6!zP~Fu;8BKr#9(9nZ1f(X?!bu+ zkK^Mq6J2Jnvh6F?KVW|yZj*0J9ubS8Tnu(9xUf(R6vc8T8^S~YqmA^fo?DP2YjS$t z3Z$HQZ24i$RL$-&a~ortGQ|w3^BB=0-I82b0vikxKXbc#85l<$z*p|s?#**Q?BzCB zjZuQi*5x9j0|n8OM`ki=awV$v+?|7y3|Ix`HZr%q^i@Q3G!zB{&EvUn<^~VXxBgN7 z%3CCbY~#eXktaOK9o45AGFeq7_Th)pf(P^moZsSZe#^J6=Dn8(W#d|uwYZ|1PP=Kv+q!1m86GMrUchd`p#5 z{nt|w0o+PrgNPgG3Ochg8M~&O?hc&u(|rDuN}?%7RWBlfG9@m49?$cA%gch=m^FkY zWJ_d<4&68(oe`a($D^Wy)l0XH&Yc|bCHLMS9OJRDXuV+kG8sk$t`ATFmj&vzjUVwx ztjz9TF7m>)|GDPG4m-ZHA^WCA&ldPEmIFsS;N6H#RPc(^Tz@rhZGMDikB#Tas(72B z#)Ug}PH>Wab%aOeQvn$TezG^{>UfcA_vD->ywkR{P&8$lau%+K8z)yy*O+$#irT_w zZi{z$s|TL`UP=!U-)GS?q_|xQJ9&FmcJLmNAXvC~7Y`anGRKWBI?_{cw)J&kGYT@f zM1LRLa`_+CryfuYdX@9e{>Y$0t`K%`wLU+CUI~0_HgSo?m!gxUW*ediV-y}s6#Hu8wwGc-KCxG@DC_nkQ|+CM~L!h ze47M4g4tz!#RsW7bOzhN9BJbt>@*jST>mL)5ON;Duk z)X3rc!dnR3JEO`9ylzTJ?#u+GEuK3;n%Mr!pl(lmRfWq$33XB2BPTWnE%0+WqJ#r+ zrNSUdW%=4pIsPjnl~XfzX9Xv5e@{tyC>JOa-k*(kIbEAxj))IR&y%`in6Y_hDWrm-fEO z&$k~Bd2=DSJ|JS^p!mQPUx`5ZXcz4vYu?l&DnlQM{b z`t-fj0+$HzOQ_vWpku5O7J7_ZB{T`7=)eSd5Km{@TeD8+Msh2N>DI)^?aCR3;*MCZ zwLGM42M8^ZHwS)Pb&<{Okaz%*_BJRBF3(xq73BW%+blhu3|~_4=tGROM9+SZtL^*3 z{#{f?TcbmFExZ9K#>mud?%X69ZtL#YSj&VHcbpS0)%OxCf3GHu)=H*(i1`wKQ7fCsN<3aB7jN#)(pg3^A zpAY}HqzJq}Rqh~PR1SZp>o{DXC9I2RszWKQ1+mNYi?f`k7#wsI)C`2c`UzlTLP8QpAZ8$Kkza@v{ z#jVOEv-qQ2Eeqc9J>k|04Fk*|XQ8>Cq`LQxO%W8IDka*~9OFYUjMdJ%6z?*b30@Kn z-%I$N(rNYe=!nkZ^2l0gFxE z%-(uNF=Rx?c)f0Fq%myU z`gsFlu3hrnti|h@7Dl0F^cg(~%i>voSk;npl} z0U237DUr2oP2VL)0yZocR43TgMgS$vQ

G27|WcZF8WMJKes&)_P{cBhNif)0Nvd z`$U7jm8J+N&-aF}NBeKtAZb+=4IVoCW=+1aFqhK)jdS#N4WIAjo-zQ#DQkgp4qdL} z!}&vUBU-fd0=E1{($Zta=L|>{eOzk=0f&DXvddl5#oR?Ui(lE5FC=~9R0us9bzV>N z?*Wx7_ggO`hCL4P%U5nSJ6Mn3A2uD7C#j|9FaG-~^(-8V2KQbdRmO5N(wR&h&i~A@ zUlVw60yj9Gp^Np0OD(L1QgFYKHAFmwWe!rhv17C~rF$B#>SE&Jx&lXiJHGKZF>wnv z&g(|ZP3 zmK2i3q6}|HIhOi-fbh<{O(l##!uw7v z2C>p^lAqD;(TLIlO*@v}vTNN5K*WRzEvb+@P=(DTg)tG~e#tL=7<+Srh zNF;*|ByMXf7u_Wx=gSomy6C-{HhPvji8?PElKZZ0GTkCpMTR97Sr4_r$(~aohJ)Q8 z%S7u@coOK@KVjnU=lhH3Z*xF)Y z_j>=!SKrvb?^rWJ^`Sw&#}T_4rdrE^hMFCFIyUUG`&9GV)cQZ9Y_+XSL`MvKfO^8) z^~|GLx43vA-8WM`SBNJ3QRCjza2`9*TuwDlqFnB?5>L>z!wfb&vharA^90_A+CnLw zlG`s6np;_=ZtFXD#Y7_`v3g)mdcyGtHgyiu0JXO?Bf81i{a7n{<+@T`z}H2#?Y6e9 z?vryi3MAv0MVvqp*4Nf`!CUc!{^JuK$guvj{usNX^o4)@YvCTo z_rKojC6sMm^zC zk0R{tGEY=EUH*Dv>H}(rbm5l++1?pJRTq^bP=*hZ=SFT;+vA0Hi-(G8#bwa@R@`uh zpYXGG+j9CHDFkjvUDcioGi^VX(?q)E2{$U2b77M+R6J|~v(ojJ%0vAwNb%9{M%h(~ zw2+2CA?KLDG-kTm*G|&cRf|2>+yTk02B^S9Jgk+k%Y2-VUZ}uwRd1eV)?P48zhWPq z9T1~=m%^Dk5LZr}LGd>GOI9m6L9$KLdCgs&hb6dU^1c!3w6 zASAUnsiNIEb_tt>Vk$(}Z@}MY5jXu@$E=4WJ(zRr7B|zE6uD8F|N3-$n*yn^Aki)k7c2I&Yz^te%-BW{Q^H<|Z#S$a+GSs9174B)YQ(h${ zn094wE~$+;wUhNVog=>d!Vz8LzHs@SH#m&Zn#$cO3C^SEKJ^W};a%7DV=tB(Fp&9l zTr-~_Y6y1mjQz%oDj!f7Z+b#EFejSsf)0Gqp(nKoiT&WH+Z>L6qV&}j)W$mzBrkJ0 z4?$|cO6cgzFMX;TR^atQM2ZhMIM|b0`MgB_a7Rou3^^WBAV?@Y1;tL#DX)7&h1r3s$05ODJfG5MYr5+DN^>z&?ek0MKe^FNOofy+bAg+OKzm>OCe+_OU5!cdt~2= zL1FAn)`>AQzw;i(Qr*wzen0o0-~8oa-t#`^yw2etb z`9r4lz0{?d^1GwoPPF3PrfXj*4HjBnnD)lFeHxzl+*Kd#X=vU4CHIsB z$)P_FigInVY8#znv#vB((_KPY(&c)b4V{rj;c{c{d7Yi6CRu=fhW^XA3=a#jFF zq_*(Rl?8CU;B$*)s$9nN@>+#(3r?OBi;X@Ru z)E0R9g`SCap;K1QdPwQFG3&P*%te)zLN!JP89=;$Q8dc$5hq$kb~duKc0>zrFju}= zXZ_U7HHk@jOFrgn+08rt+nb~A?j5q8x@gx;(XpDZFCP~ud}XR{_%u4=wrpZ3JOkl3hieD* zy?$~fwuBt!*jZCDkR6|dDgi>Sv1Ri@Piu*Rlm0G6C5e0se)q>_(H;@0gB^5`(4t>) zJMX+Z+{F0vy_53Y&i+Fg@J@#W<|7(Y(3$aC@sdkp$=iaC@TxarY9KFn(&K7MmQyn&*74fr( zDXM!17`G?wW7i>Sd=ejIbUpfs?VEfa;J2OJo+IH8%h-)mC;>HCcU|3byBw2XK7r1)bBsSZ_ilBe*~} zivWtAqMQ}{w4v)~@tExj*CS?5aq;`7)yrRKbbHT%huk3dW6Sl~770HA-$V{xWR2?V zWg}o){Rxc5g0FjS!j*jL?-RS%(iLS%;hPw#J^OwJZdHZt_LE^=W~Q_1`j=&IjV98+ z`p66Lg)zb<3rZt8r(|lJQOrWmmJ|g^6qZIQGf~XxeO2jZ$5fw%k|Z7@xKAO5v^FAx zq#HcLpKBz0bYvSPnfBF3eO*QD}_{keFCZ$+_B^(t@}pYIVM(nDi{%*)i4c8 z^GwD(ZfX7jm-nd$mi2Hf+{K8AAu@lZcyhPu!{ zGX%n}{It0^)bun{?n7@I?pA8P&lOjC_T^gK=bbpT4Fe?LT5wgj%ZAZCZiBK6g%k>g zqm4rNJj%$vJSXfiq-ejBlHI4RJUziQNvl%+I$AkA5|!hPfE=ZbOIzS&ktdZa&T!C?tew{6!PGGQ z;Kj9>sYMUG;s|poT-Q75-YVB1u{0V-+uRo-;WZOeFuHS?ykQ92#&Bfj8qMAdY7tAF zLood|nK)6s;qA9H36_-XHj#YiyIFPQ+@ZJ}{C89Qa@71z&&#?O3A1OThi>=JfjXi0 z0SeJtRT^7?2$Q9bc6E8F$Jh%cqmSD8e|%u3{!`SRT=*uBeTP7q)p3v0s>gMd68q8x zi6hkEi7*1NGYqln_uBDi%ith&7EoQVhZ9d7{s`)3?nhN+WJe<9@Y5pSXmpIS3~efwW!9 zrCSrYT{BpXwo#PAt)BPxpkx~|4tfcLPIhTgS8eFD$D`-q4`vFXCO7vKR;b{*o7A2V z-+3qFi)nv5{~#k+M<(B*Sfj2s`y+6CCyMg9es`+}GFt7czO6_%wQg$EvZ>{ZiJ0HT zq4aV^J#}F>$|}ALNJ-S^s`MWlCT9!y^}S(PXS5fcUtcpij2N+uZP^N&Yh14@jb5D` z-H}HtP8$9CshqMs8uR*8pK?yxtuOTs4N|y^P;GLp3@dM_$Ba24-9BGo^jRCZMUhVz z_t9RF2g$Pv=FVl8>Jb=vG|DuW8e7&NS~$h>ZGxlK^24BzW`{%h3Uyo5 zSb~X`pj)Ry3ekPrDEgrt&o7NPcV2oJg~66{;kxXT&hI4K`QaFq=6FUT#k%K+)?rpR zC0hD3PG85BZdBUkg|c>bP_Xm&-&YCCFoRAr%!nbu0!;G7oC38zyIuKfKOQ6@G)E@7 z6Y{}{%q7uR=Eh*Wu8Ac?lsrEPSW@Dg?5g(V>?TU|g0eo#LsRI#CLZ_e=+F_>Xk0OK z11LHHVd>rEeLE699wF@h+!I_N_d$%bBsLDzifijkbmXobEUsZE8RkY$w9_qGl{VX;c}azsbC5OOWSm!&I^tccZ| z>RlQ`X=NI zv}`|})J&B#n#vRNesYfGmZ_I?c{zpbR@{+OhQL>>f=z%<6rebf7s73-R)%sA-DB!* zbCJ{<`s47?V3}a9{`Rc{$i3jRIxnafyECmx~k%-@8XKAx6HkGBw_}WcBE1BO&ztdZP{WEGCkgS^zocW&t z*Z`R}HQv2F{?)aSO$}ME(yoa1J%?H*K&kKRi@VaBJp($-YZL{X0DjYi#{#i!I)dfYST#A}g)_d+#w3VJJRo&PS>eH0xBq6W3y&W^o|ShTwjp}WC-CWlz{cbA zsrw4*dH(v4({dbiFaQ49{PRhE@q1rENA;}WvDulo>et^)^i@=1L3_7jC?U%Z#iz;I zrZV!hSrbJdh>&r#tG--u-5J3P>tRn^p$=PEe+Vl(xjb%r`Fx~OJ8`NRoa@=v7^2I| z0mFh3c^YTxSGYx~Ss2if2OZ}bgl%w@SxOm9(M~;25%a5k_9T~i#vkld)`Ch&GLbH0C3JN`SLtwjX+WVj3O`s$))cRL zn3sfqZ&eyi(Vq&3_40qWmF<$)+FopFD7plB8pMC}NpiZe&Urzfl85WIw8khQ<5 z8JPtt@y=k{(fT-b1NT=b=jq`AK}9Dpb~i-Dk6^Xe`cNFdEuoq?{+{B0UOmZ;`kUO2 zsvJDHlG!pHiS2!w5pV4avL`7VEWWVVvkY$Ie5eZ zN1VBKadHr~NDw=bJKJeD=AWz(S5$xXfuU8-bKGF#iwumvKQss&8l<9yYywo$!e$O< z^w)QJtqB02Hq1T*sNY^Vl9r_&8UlGY7N3;;dUSn0?RG)6`IX}xqfjO^$aCG9wm}M1 z`x68J`?gk8>dn4+;NpZNx>55js^iB8thDb|UqGonwK6P_1SJ$5qp}ob67yb&zi%zI zllfWsl6}YAtsws@>3G*K_nA%MP~`;WWsbiI2pz8B;Chyya}45$jnJ;>yIMxrv${lC zVjr)e&#S)6n?#r22K5ES34X^9nisHt8kt|Y+tynoNW%wN4t8Yj!6g?1UVIgwo6&Fl zf!rMaIQ1+iJmX@lf<_SMzI&h?sqkkE!)}lI6ydt9_T5%=Q9C`A`S>;G#YRXhlwH2s zw0g9CE{*qY11H--NAl44p=F$6fWBf|{JFe6o2SliJ?voqg8j1;hj$9$Gx?Kv`EJz6 zWpeJhHao>x)=7rdE;}(00oZ1?L z;QP&vdv?4%PLb1&NNH#UDbdMlt{Ix8n0G=1t{IN#fh@^w{=Y4#WmRd{=X>Nun+egOXPe~>8AWvpG?3`$YEq(FdUyF69Sig@1WM62UCyvT+n*10iBgVuKUaYi^3w$`>FYfxNFYeZmM7mPrjC zYBoFwTFa_Hwx{G8s3L~c{bp_rBlGg=L@Zbu<}TP3j1jQoU{>7GbCf%WW&;Ya&v7g#Bu@=EpBw(bBTZ_dy1MB}EV~-pV-;)6iE9EtJ$McBswn^* z?0ul%5WI8V5+W79{>Cz6jn|Q!ee{I5TEk#ozfO-874)$jytG`W?28CmnGTyDXE3J! zmQHxafzaj>Zn&D$bxuquRUPL3?us_3VN+>g<%ea*lE2B^L1cM?c})PmS$Vr2)M58z z!*}ajUh)fN=G;@Bl$3A>tByJ*W~%}4?q#OCN9ZlRZ_;Pb5Drzz;(oPzLvwv^7$`H= zzBu`FTT!z_dSS{Co z{Shq87B#7(6W(*6O;DpA9gvG0^ai6Tto)&BL+gAoHY+cgd}IVT6afOGM}FJa1B=40 z+?)q-ne$mn*0y|!10^t)0lA8Ue}dQU`cJRjAo~(8><~Vh73L@516qKM7dMXW$g)QF zG3A4+@A4vz)E2I9N5TI3x2^;c~= zo&Jg=#!|b1OMP@!Q&k>-vwvMWI#;|m%hZ5Zy-}l{-#-hm*Vy09vP@dc@0PS(3(cXi zR9+9J0{@%Wm0Hx}g$4XcXaV1gpexXD>1bnjjj`xxR{D%ry?;CdR)~z`J?B$3>dHY= zm8L)6gCARBa5kMlo7Wm4mL>!k8of1p7lfW5{Zy8BRAwVfkz93I68@!bs!{j#tZ!F? zYZIO$HW|!mosZI;k9={kRKU=&EtXH6HBPEF=<~hWzD%bs>yg`ig7qy>9gNJJqgT~+ zvJZCIWc)fRF;zsjD4X8;8V@xG8sh!@6uq{V93rkV#jZ&YRD!F3lc#wOSV2W@&B zs)6$EC)m#b`PFa03rqNsOcL8-H`@O%Z-{!w(ir27;&u1l{4cA}s#iA*rG^!wWUfPu|YZn}w;;Yt$Y7n9F zc`X=SVx{Br$ol(yVoQ1X8{7p3hmMDDUR?5U2Icpyd%QOjvjVS8UvO3PIsss}nqm)HOB ze}aw4n(G9dy8_)^_ zkwm%8IU|_|yXMTc{o8qNkFOH*JGwQ;QWKHXVN&>?cZBV!uWZ~AVN(el?jdP)CEWLg zv)e)NQ(CLwanA5Y;+yeImN@GBy?Z77$N!y_X7y^2KAv@zklRm~wgAn371SDlP?M6E zV$0g^icZOjt3Frrmh{~}+wj_Z+HiFR0~c#EG4I_WE@TSaL`qEx^?%)Jvj&;EfrPNe zHl2UPet|%znUv)*u^rTzcl8(a5wCJ5aOH z7OGXzGMla(rT5!9%W1H&QeiJa4Nx@`gyefX^E>u;rIL#~?t%1bU5_Y^D1w*r9s|fS zi>BE3Ed4}$CV2Ql9ej&^X^jAf7ziNl&MHkSXTVKQ?yHRVZ>|(!`*VVL5Y{|SdLqlBh(e{C_)F9vR zYw67wj920E-0i#{5nI?307W)c#y9!tLz1lx`;jZh?pj-3ZPpaybt6{d&;G&~FMr%D z=d=T#<_==z&4O_%{?+!_?UKta9?5ZFxkfzhfV@g2vo1ciH z40V1f4VoCzq$4|P+J;^^Je?W8-=1=X`dg%a_Q!AxpD49t#0WQfMjkVIGvlm8&NHG# zONxwZ_XM$t*Z-y4?Y5hn5cds+t4}>*SWY(gxANu+)k{{zW*m}e9(Zad)hduKa`nAQ zC{}iTeq}KPNXRM}3AXFCgW|dtEHpz8OSIEzoZvLE_7!2x1HxtTei3Vrg637Ib|p-N z%T5Tf=lBf8VE6c2O(lz(lT$(6Hl0a0tLNMEaue#NRa<_S2W1~EpeajRzN9-}Sx)uk zl=0-j%;viO?J~^`V+yZK_Y|jfbPl=JiT?@wvDV2DQvP%^^`6EY9%V6PK~H&Dld3%B zlt*$roJz5u&5v<`E4Yq~#sWm3L8^P+smmGpZr5Mr3b+~t>*xE8W7fAvUK#J!>Xp!M z8?Xgvf9DSd*+>Ki!5tL|j z))H+#XVN_*b1a$|16)zHvjecpQ=dY)5}Np3F34#reTrKXNETYpzxjPUnKIxWp4G6b zbllin>C;o)OKlRoGiIEbJvv!R{en*Y;&qf7SE#_njkJW95(`~&i(}b`&Ik#P@ z6ttVTV|icGRL?{ZQPWHwdEd&SLk~BYcM1JD%kBE+@a(zp<9!V#Ic0~E7!nusI#LHI zLNl1>gXjZD)6+}|Z0MC1S&JZ|g-d4=W8~8<6}cleCU+N&+O^z?8_AsXL$cb6;=UHz zcMab}RaDk+at87$_E1{H#cHRqp@Wu&6sIXyfPA=Nbkgp)+Ej+iM?z;)gB|wDUFUOg zFYS!pQ=gjEThelIslyqiLLM z*`pF;6}-Xu`h6Gobf=t|SqGy^N<;4f-1%Y{i&F1E=hgiwD}Ay>XqfXx=}KQANf&@i z^q2Up@*vJ%#!q-$J~Et1Gd1cQvZ$M>w3Ib0P@q?9xhYF}Kh%cv|AB5BqW-pH+W7(R zd?h<~l?e_o`wiO8H$UB8-wkRNAZOE-Q%GkgNwR~*aoq-6wX@x$ zM#m>i{Qm41iJsMl47`RJR=WC>Iffs>KMFxG^O12S=mwfEoA?FFNW&OOm_LyJ~)+)yAM^( zxt}mhQuv}TkWfhyxMpQl1^9go!08< zSQi;`I+EkM^5E@KBZUt9SUg>XhYvS)%ZM9C@E?d?=UqlC{A#~nrkNaooYWF$hzs~B zPtiKO$I%*3A$pp`<_T2ee<}`-5H*U`hp+mhiPD!fKC48`N|oT-{up4AUTL%qjxP~# zQmB|193qrSCFOSR)J7iFRx%}1`^0^TxbeYAeDn=}9ua=gqsoQN{YQPBUGx>zq2#FF z8;Gr%j0pCiOVFS`SGu_9NJ16pSfD0x3aFXanY6O-pKw(&J*u^R@v4>LPDl_l?e3Ax6RGbD6VVA&ZLNVEygc`#^xhKDv* zyK&aj=-9!p5lVy;_|=mV8)q_XCA_t;pYPOLion(TXQ#N~vjv8?YD)pY6)m-FFK+U_ z)djb6lh+vxX?FMPt#eX_iFlz1zhNP-15k4@%=2FSnp<+)6V~9!x8=a{$8nmVo=n^Z zohI2}_~Eg2tYL&CPEv5V0UZ<^o847qeOq)i>ZYFJ!;1nxC3oC&A#mLyc!UG&RBKkm z6xfIcJd!GUpkDvg`j~kSpjIr{ch>2s>yQNp0Q7ML4wKm7_`EH?EvfQ1>!bsMspGT> zJ+v6ZTKUT%dW3fHzUlO1%gjA3*XT$9fza@ncW)pFfSg}%a53sMGtb=^Wasjp#Dqx5 z8kx4sp8%Mm?c~)Pki-J&5T?@tE|84LpDj3#;XI&&%}tCaS5{M*Q`Wb-6FoQ2=5}0( zbt*8gV)r$Dy_yxab3piH)<#Cws(Bk^qb3K*;gbf(`1sqxa~yOBSc-y(z6M9HgYiMZ zavr1}HY$yz7as(^LAfe3DF>*;!tv=Q5BFpIUhg^74~MH~8wK}qRrDgbN0%|~ zpqz#^Ic7a(Hkl1Caaede@;kSQJ?nWdPU`ZMpqPN~y0TB_vQ@XQNVc$!=Y~mOubF}b zQMHe;!|i!_U2^G-O;>tcqAltk-8OM$u^WlC-CVtvxTrC4sn zN$m(y>a@z;*Z-DU{GR)!CVs+YUI$$&^PJ|)N`U<@8!4Y?4+^Cs`coY(8_~;@!KZ)X z=C9;e^Kz@kaP5fU)HuW{;7Tl&#Ec>ez6pc_TqBkT#14J{GvMjx72%m-)5}}J{l=f+ zKECTOOu(-s+5V8`}tv z{^|w_b=&^;_xKf6pArlWfUgP!SVR)O1oy>+!D85!{}Y#&2snEFp2;1<{8X`fR5Imb zvERp2e-PgLR(j>FT2!%8Dw)^nj@9KYd95qrt1F{J-%{1GFsJUYHfpQZh_@N=pV*~6 z>`3wt0-fIy^vHl#Abg!pgZae0|Ahou=1%%vc}Jr{Iy6pD4oFEZG05_0uNUy-*M z$M+kkA@^9>tu~wU4q_GfpAwNgRcn@f*dIlCVUSkX!)kgx(!PW2xw^?u^|jxR=e)UY zk_CGG7o?TEAgH%2%V5>`*gWH?;Xg7Chp<4$Vg1YNJNsc6mAkAtv9d#`n|J~L3ePnr zvWVAS-wy(X^ucw&fAT8pz&G~@_-5R?`grNEVUwSgAb`O%8p=%iuN_Y0{5ep`^rT?6 zl<4SMZU(ZE_nL}NVytBtdW9lUdgbOA7p^pcupGYu~kX z-Rkq(cc9(tG;5aZuVA5B;!X;7_)U#dbFMf0Lqeg(CdlW?U7?7V`{6F~PnyvOT z|0)|!?n$Yg!K#)lky`wIdlCTC9n4A)|K>Ls%O2jn3Gp!_+6}CIS^NQ?t=r_e9^QQk zaq#^sryxCeyIlOd76zfg%8D%h!i*FYU>Fq@;peE@&C(GW7+$%|Qfea}9J}Sa1OFfpAfP$%w zHw2(W7&ZN&lfHa{UK#CGixB-#ic1~D`lL>dIyhX)m`@}c|KX2tN>A|Ip#1T>JQeok1Tf=sDL&&;{Wmn# z<%Q>#hcf5^2_ZGKZe1mcrg8=m>t%CDfX5pXDK!kj^5JpUHKUnKu zc^~pA#*toexzd*xp1N7(1R_oPYCNC)Hf>=kRJr024pzP|H>B;MS6tGvM?t|v;~MQD z(%5lUdp+X*8@fA7o9f%|+*0QdbWe1CRQH!J47ZNyPAE>DbFgaU!?=L=fDZDyJyy!Q zS-QQ6bnh1x8VinaVGBNl0H_8@--tw))Zx~IZKGeKmgo1NkyPkShvU5#5`M@k%LG_P z4FpBCe9UywKS8+?p&q-l&u&eoUMflp3 zam?Q|z^Y8cbJ@~vFMA!J*8$iF3nLO#WI!Kd3pM1{0Hk2fNE2wo)$|FHS4voRLz076 z+ooleZeVM)R)M+-fRv{{|9hC>>pBFtog1hl?UUyp=sr?f$}gz1G-EMnTbErU1`)8d z3dsQ_^+vN2mTj^rFX9h%Aq)zAU7Gj55<$jlRL%=Z(q|){e_mV)0B|^w;!T=qO79Od zI}6xl8R=tAlH8(}R`wCF!VMA$&4R16!Y|v8n`rI3!(I`7ByxgijE4Pxqi3S6eM>Zn zfA>Ob^NRVDri%y}dXr3EfUNJEZjWVTU@=zC%zk$IF)3$b)0>Mc3(#zBG4(be zbqsydWV$bJ-6I#7s{rwJ#-1Nhhw)&+k{kvs0#PS0cg;e*@Jcejfg{SN`(KnJQwaO` zp8jaJaqQGIwA;R+xm@H?*n;DjE52;w8;WDRhW*NsS#n!^i+`}`RmyD7Z(-B#+hN=n z5c=AAmkE%$o%UVG5g?(%d>qwI=KLEL1<)6CD*x52K=Z7lYfs8q^SZmd znt>8{zmd7Oq9al#*y1gB_-5U&ka+%6+G~X3g;WKD0cg5S%M1UHZ}vGQenl~dh~(AI zKDF#?|E}i6T!T1uj)x5+vB0S!7!I!mm`F=)^EZq7D+|t&oBS-nd`7#4A2$KI`>$Z| z;`@?|h%%tQ{yCRDGT0i8@wp{r*ODJcaU~Nk*)L2>x9qA!=>fS?Aj$uHQ|+~jXSCa= zHjWq)ty@B_-DT;7R5@OIDZM(Wza!g3qjd-FRuldsiRQ*U>Cupci+8k#zyL9ZI>W;A z;sDD^d5Pf8R!{Xh0?^W-Wa&SByCr;6QIudieT(y~UPbWB67Hh&*3Q~@M=c7hFi!fc zW&kh_xcFjzpbvynIHK4SxpPWWH5(!Cl;1Em(&^1`oqYdPqnUD_LB?G~LlmmM0a~nY z$qC7kZf{bwtkG;VB_9>T+j@TtjJ!TIVSKWN&RXsZQFFJ$0wHK;=6z;yM5LK02&!I< zG^}JPL?9c+r~{Gf2YD5fF|(!>oB|!=ln!&v_}4685+HaCMYhIAS^$uGUEb3QoB5Bu z&Y{Q~xl|=W^mWt|g(LbcvNu}hkJ~!NH<$%=r-_yRX4(+a*VahD?`5&o1${mfiNI1G z|A|&mQa*O{CHaU>cW=tc9Qj&DEcaW?pd^;rqu&5|b`_K3~f-8zGUD z=X0q2d^if+^;j7Sw-!*~9aWflCux+zNJ*`n&}_K6k%bG9jVQSW>T{uw@z5YIVwb<4 zZ>;Q;Q%&*O6mVn3%U7pPWDhlHhdIRc;mSDm_=O$7onWmX&}v4D#nDR`?<8GikSB0o zG7I!Emu6;_CWc?U?NnwANKQhKf9pD(!3AX>Mlvls)H2#RlxK!RER$S&`ujNAOlJ*R z^xH@8(UfIAOn>bo=Jh&7%njQE{_A_!*Fe@yjm@(pkQS0{*=PNSpkD3Q`SsH=Lk%;T zmT|OC)rawHYFA7Q%e}m`_cJ*8dm{2~;*<+S_g_0Zv0PC)i`L*Q}9eZ3t47H>xc#gLZU^n;MRhDd+mD}V; z;CBrKCa`&{TAQAxC?za|M2H!5Fq?2-hx4Po-P75v`h}&VuXclE{c(e$Rdw-TGP6MIyZW{WG_jIzG9V zVeeZyD<>Ye?3cqu^h>4^^SCr7`zWGFBUIj}eRG*_j%R3~_C`^L#wRRAQ$;0@)}nhY zr>^HjMLX@x_;TE#!!G&+>Y*0WMZ1GCPHP$;X^3oEyBQmOa!O7Y1u*R>1o0aYO0&;) z-6zj+pi}NM-+@GI6XT)%geKBFFsJs$tbgZ>f&odZ8a|j|7oakw{6N2N51OH%e+fk_ z2<|(a#C(f1>Km$!oT23?1izTDh&U3d9~@buj{?nv07wtaCcZk!QWq_732d8OMUGy$ zX0Xi7YNy7s98Ms(UA6{`f*UoWc4_P9j;F`TC2Or}1J}H^;y_X+3HQRSMF9ljst}G8 zQ6lZ+lv^-0XLN36+I-AYKEDc-$9JvU!p;6GbFHHCwWZnTW8d|$Da3u~)_4bA?&KtVi1gcu1& zoWj!4zQ7jZRGq(F4zKde$$5OFQ63%G8JS&S`=n1fA$Q~`Gj|qjycOexXwbwRG`LXC z+NfaScsnU6F*H>yHFZ`nyK?-F*EkZMNM6fDcDWIuEN(raml*0_1?=$z`{JGL1)LDm z3&dy68&lwjYPAYPaHHifc0ARc%CIwT4kwN}Z2aPqdyg!E`9Ocyw6V~oW00EE)J_>! z3htELIVIN$0MQOu?Pe2#Tf5sevxOGQOl_fQ0$@GQ{7-)R(7lIAL64?l&y5CCPJKFG z(}Wxq!LO+py6DfV97wtQu=Cd#f|>C}gHKVFJ4%reyr%tGk3lorEDGJGJxW#6fuHXw zyD3a{<`pfw62Xi*pE)kYM=sFn1p-)LD1iyYmfMADAAp~~Y0CB5C74bi zky9g{<}q^nwQ+b#daGETS2<#Lz5%`^KnJ8$O zj>1MbfP4OG)sbd2>o)9d#e6HJtQ06?_p>k#7g&}9omVT_etd7fp=uN*iNJsJgU}uA zb&dcyjb`AYxGoQCXFdMHk%seEE*7pXiRxKngg3s}Nhw^Zhrxhf7szDqf1-UJ3xXC{8a@YQ3@M$;u-asU#z;dSGam$5#E~KL#@@5p#;y7aF$v{Ex zgbv)KKSPXI=~18`Vbt8U#5&xWO~A?>+TI z`}AZsh4dW?>ODMBW?-iDig3_45PM43D!wwwP3Qjk7MDa2(V&orTJp>yjGD0HmX*}Z zJV`RW7CVkRsNZxjb9=i5{4W=bpj+BaS&3jPymuKWDTO|NZGdjVm^qV*p(HFZoQy^a z&^8-}o}YFWUtP{K6NBIYHzDS!+KBYGYuosRXmAbDb?C<(zv}vo-o` z+ipak$ayJY#3?!69ZL-EtD`qy$AviSEE}nboV<23PUEh1Qv>0@k{k_sv`|@QZ#}-f z5b=qku;iTtJK(e~C~`u^Clz~7Y>JM#)dl-#bg)c?!XHm1_Di~zgCnzdrWzcO3JF%~ zSf%zOR4!&~mJ>h0C?u)MUlV;^x--R2LO=U#XHqClH>9(wyF&|-vN*txC;SKdzRSN2Y5b7FOC4Sd6h01J0kYh|{~=4T-PHPrKL~~WkPnuw zwf;f$>%UL!+j!LvcYNu(dk{$3=044WC0~f1`~#|e6BGQg2rqT=eIJDTvD_|(0fF4^ z;nvV!blRZ-cC&gEps17ztJ45X{DB!^DGmiuhINk+D~8C5djg7|V%?mxOyI#kKobL` gCpOn*pwpPlHPvAkFT;@Oh0sH%5$9D?P8i?$f5Cby4gdfE literal 0 HcmV?d00001 diff --git a/examples/img/graphical_pipeline_simplified.png b/examples/img/graphical_pipeline_simplified.png new file mode 100644 index 0000000000000000000000000000000000000000..12a50c633e0b17bd01c377fa1378b6839519f1eb GIT binary patch literal 80681 zcmbrm1yqz@*FJoYql^KA0fM9j2uMi}ts;Vyba%thEexYb7=VhvD2)su-AIRkgml*s zl0(DL@xKRu&-1?T^S$4Ht#5s-rQG9vVxN8X+56hpKI0o@MH#Y-G#3E?Ad`LiSOow` zP5|J{+w&v<0715RJO*DtTvTKp0@>|!3*d*dmQo5*0FW1U>CpHb`1yk4Q*9Rjpt?f* z18HI2lmLJetn6bcHBW=3Bfljv%`bCrd{d^o@~?Gl{1*5^kk{(Jz4YFaq?;kdGO7b`x<>emSf6a?3`K5f&%B{?`r(z@_I%gq@t8X>5`sKb!roI_f8GUu{xc;Wj%}dXo z+&mj7@s;sR3HLAO`?;f9gLTyPJJEB#)pI{kD?gORbW2d9y6QWG?wzAb^|=^?Tf5g{ zZ~C0eViQqbK+?nDAm?OWjyB>CkkuvSGXIh{70Qu;Tb?Q5U)00BIQ&w{)@*FkYuRF& zwp7L;_@(f7d{#dvC#r92ggL-h8UhKBtuEFn|GG`CKoA$pNl$bh^LERlx9P~N?<4lC zSTMy^ILo`uqY?LtLOi}oa8!GI#szox?O7VvZ3Tx~Jd8$;Q8W zB$KarJAds~@3zq(w5g?r&S&{j{Rg+J0|gyN&P+0`hMx1y`P-sOM@v-Z3pPzhjo*}V zP>hGiN&*|n#eedcFwQ=E@Gd;6P(c{K1MP2|@gA^{rr{af7r**u36 z+I5=_0iuv*v1ao0j@+urkNn(6o@s~fGh7M^2Y!#bp7^Z7wXTPe&^wyfQek=_T;`27pN>nDBHWg>j%wx?(3$BjilYhy@q@GWNxhTXN}y%-~Dn(&Q@37 zyXWmH>C|5=BMV? z{@cR0PlT^7%V-U2eP;trjRB%E%vq$TS$pQPIqSL-pbW4KN7L8>#Rn-gh=g5^o}r=H zBr|Hq~rK(fGyXoY^g$m{RZR^AxxH;-_qqsQ7jVZMs>$~S}?bFAXK^gqb z;&o1T`x^F)V^>*wD$U&%erf0mUXzqz?uNu_m^VKe%2Ri*43JLtD4}~;Tn&C0qa5Sk zY|eN$Vs=2QYd#SBYlHYJ9FKnJB+{CE(agj1eR@8g?7pL=@W`iS2YDnNFy>Xl`kQ5B z4zsAsh>Sm4Hj$6eckm#gw6KcE94K7c3bFodX%T$uZvKGKXZ@T!w1AtHbY)In#8C2` z1%2Kj@g_CgzSTiyIC0<#s0HoT^JeegkZ&tt3cA*K_--e;U)esTLiK9JL3svFWc7s^ zf&FXuV+xiC{hZv0;u(j4)jQ$A=GQL{xV-$B;%!co3q{cGnIF)py4HvhEhPbp&^3Qg zvb!W0uSM_p5Wl|TyCYt)Pq}EMGTN`M+7zWvKEC(3GDk;IZ@VN;cqqS{ZXGL+7sVfi zWECu-%gFgIo%~E)8M(-=owLvh3BaYF9OOaYdJ7*=9y9ZZNu-9Rn`>TU$T>6Ur{7W; zr4_29Bzm;jmlGxJww90EEMm-ha>6LO-ew~lXCzu#N1)|bfl*}m;E!Yt;&}KhV1@G$ z`K97I%UVP;D_peOC%BPsx*ehRkGULP%V_V~z8c@yK^+NFY)2;TkXWDRRK>BF!)76-JdGiW5!!Rk%xXuEh$#*g@&NCe``E@g<~XK$RfJl7Vj+f&~M%Zq4E zr3^^pfq(M!IHSgg&l)=q*+Zrlee@!?*^r2YlP3*|3mbI}+|Av{PN%aJ8NPAfp#~gv zj-GKt$Bk;jM!HUtm3}*qG<#AG-aF?D`7Pu|_6hS(9e9Vs)0G-*$_3~<)eR?E!*=nl z>*ezL>6Pj6wX34uGYSsABa0uumEhbD*hIxq$)4n4WnVWpU6;_qvpO6xNzlkM zdOv)e)=fCbWs9rzyI-b1Q=pGQY5N90Sww%O)x-YY7EbYYt)UEZtw~gKudIf`6T$Nh z6-nTfDQUsnFo}=3Zy(>1?KaSAnj+J(c+>1)gry6+DSO9N66;mlCyg+ojnG%Bf~G8# zxl78LO&)-QJp7h$t}v_KApdAN9P<=5`nCg{q0(=l7 zQ{2HL4HNwk=*yO=uH9?Yq%q|7R=SwY;~F@c#t^}PX0h1i0BoJHRCY}p1^)7WK)*I` zPEC_hvxNWfyzd#;s=k?+(vz_p`QyWyOn83d`2D{49IBV%T9Y_+BQos)emI z7{Q5U2%ib2SLQO(;TSm3W7BB9fF^WjYy(ri(9IeNoDuPHjepP;@Bx95tI6zFC zdio~#y|3_DqJId4@4EH$e^yrgVtjKR-88e&(BP6{VEAh}BN=LQG?2q-Z*s-bHsjj| zx0*!td;^zudSxwG1rx)dHp>mw2nF}y+mY9vOi;v5j*29F^B-|F?U!v-n@7J`PIwS0 zO$|j%EOgd#n8zk(JigALjxb`UP4h9o(ujt8dgm)%ow3r+`F45Tc(jyLG`8*#EA#w| zZs%EgN3-4=ii!CzxEJA9u_dn-`d*Lc4JWvI>n9rYKZBhed^9H2c%A<8%@1y$Kr1+V zo*0F?AdoXekz6IwXv9yV3#LNcthtZ~Jri^J|LuW+oQ9@2;k3oXm0v5LsX@2Q?KF?R z^cpV2IFFV|VMq^D2CeowNe>u7&3c9NH7{+0o|KHanWMrN_LHLlmLXV-Or6PNe%|=cnvKuFTgC zHRsQFN{&H-z*)%XFDF3$k@P;@KM!YnZ-r*%+Y-?i09Yx_8t|vvo#}i#qJXd(rW2AI ziA<*pHmEj8Gc{LrD4IuKV~V>KmY#p$+q+Ph4c*bywC&imoEdn2=n(z+cz{Bg`cHzx z6u86n=6PhOZe_2Pd(|H`X4G^o=(GDFNO;rZz~(9=KF}pdu1BSm)%W2aj`WW-!L>Z5 z{Bz~IX`7hv!Q>HU<)B?}xXBp5c&=NlK6Dq`*KXu#hAHzGa@$oN_TXNc{D2Je*3)h} z(9@1#a;>}8*$W{Y3^|B9&E&Df>jy1FukIUE^dF6^9v_6r#8)|Y44?{iJSbL1WC%-W zpw(CtwuM88_8hGEdRE4`mEA+!l)QyUo4$jWO63$Q6X!ge@|rXp+aS=I!t>!6^K z#+X1=4u-TS++mKH|D&Y@f2E#_`vhe4@x~sNCY7eu#fQcY_p0kkJbW%n?!P^Ak&C~< z6)9v!^KciX|51X!!7;+}-8g;j{ko5NffA+ACydJTVSL)7;sO4N8ApB1?URGeiVtYh zI7YtPZHT(o^caOy*?Zhz#@|)g2%W9=@LkUfAqaejlZ3uheeO{+s2il#%s^`UK1-kj zA)7MTJhhQ1XZI>ZQK}zOFk-bI4Br*)zpRUNdJ3bHBNM1qYzl;Xie_HvcE>2gCK}$#8v88Rnx9${ZKR!E3@1(mgC(LKxmtbpH{n3DpUKzojxhZD8 zZ>Al4Cx6IIlggN4e-|C;!)_-7qv*i2C*9Fu;%_Q>?Z3m?6QgvyXZEq*Yi(B+wK?{8 z&n5+Wl928X6nh~Jt6#5Ms%3PlkoVfkKg$|f_=T9DW*f4+w=$dYZnKfi84jBl?p^on zrC)p8cD&@x9NUq24c_n=k?<{F)Y&cXi`{Fnz?PMg++YJFW*JY-R=z4Xy**R~dVV*S z{KL@1zIK*=NNlXR%49*gNMmnVE!+KEj{}8taoDF>r?>l`^*tWW{62Q)N$FpffH3A> zK!)3=9zMA9K0OnDy%TGoD~}bYA8MJSAh4FNyl#mGEv?A@B=ZzUDVeVFIr`j|oX0-y z5Z`HKCqG(C&DcpgU&${0bL;q^y}Y_yC)8TcNvwYoJZRBKtWvB%Fu|$8_rN*Yuup$p zSnA63H}BzOi#s`);-}$>&B1*%G~Ml>BLVuh1Jm#I7~16D+Lh0Ut`qr^x=wX{L9?TK zfy0C+ggyHOJ*a(qZzJ^&FTpYQs*VH$-y!ZF@8_=(bfAODPxka5R0?Nw4C+KA-1*nA z?G4lC}&TmS24>WoEeYw*3dD)fE z9JVz07WqSI8z*J6h(2a7%)IvQo16Zd`Q%9!idbdJ>o}Lid*W>D9=agXTED(ua_^5n30b%c%iOk(p#ILlE?TYHXQ>{1y)er|Hfp#@B&1rr z9@|biUAfuU%BiG@x4D2=Hy{emw2F?jC*YgB+&eZ?S5G#c1dUB2XF|`>x=H0u~rg4M!&XS6|fY{?acyVWmI4zw{yf=lD$)GsBNq8hdKXt(* zzKZSC(EfJ1wu&|sL?TBTYJwONiWU)rasvpYla=ox{2skd^&`9Pq-@%C{iPmFFqcsF zDwuYc&l-^c9b}rB-}#V{d(~~{P1!R1&$*@#Tvtt3x+b!?ccn*Tlyynij-;-hku_E5QyoNocu%>W#p%=r78~{ zUGIkwZpXAQ|2=rPXDTJJBNhiVe-j&#Kn;=jr!ch}DaxS+TIVbbgWi8Pc$Ycdj6su^ z-t*TnzwUt!L~vx=v#{XNL;SXk{H~G#M|*t@=Ejt|y)ShZ3(^n@f!Ew>8fbgiN-+q| zbk6YUU-Zh=-u3<3xoD^dL7?zluW4D~GP;aiuOy~H(5z`!UohdgUiV~~?Z^gZ8K|xU z%e9w`r_GPD4>jzQs^&g}6iKjuP&D7??2EJ1Ywr(#)>%Ldf|l~$YAn)!{iEH_p%(2C z2eo0)c*p0_48;yZMNNa2FZm^>VLH9yLw`0f!^YuRYbnn4K+Jp?x8*Z;ZlyG6ihD~7 zDO3t3Do%iieS{BMeJiWrzNrnPd&KnO-cNs>Wnz$0A2LhzHPJRUT7;i(dIT$Wt%Ij2 zomncXZqd4KY&DNg1^?}{Tg{{RRS`A68BLV5d-aZutot`2exd>K(@-A-Jm&{5)by$h zs+F*cRD`b{sKTPL1@xUHxF_h{{qr)Qi$`4RMX$Zxl05F85C*s#9eH|fFD#&Qt-y4) z800g~{SOTE+)^l@Wo73f?5xl&5HMZ)-yTHaHfE>rh#|<3qZ*(E0Z;w%43!ZG>3*4c z$i4?*2?pX?^D*&J4d*->L_tP$*g!CTS)2#)n+dGTdE2H|JPF!P3%2is7*l~wtryXR zAdUO`EyPSegqu94K4~7k+4w%_s4+Pmy7=O9-ktPKS;pgeecH+5A-SsK?GbyNRtY7a z%fcY~qfjAKCqKk7%hI^2?fm&%{}i~;XK-aAe*qj3>-htJvEFl-XmLw6{yibI z8$VxAE?X8Xu>++;d(^lnr*9~e#^T~5aEX(zOJ$FktMr$7BNh60^S+Q-==0S%!E^O` zjq}64aLS@Sv8pq?MM%VIiQSJAI!QXbNIsC$UH;fTpKv%;H5}+wm6O9vjcdSFZt z0S&b(o~UDX6HNDj6Y5haG+~>mPy+cad@p$&(urbZlX^7iAJg0G9vazake`(rw*Ic3 zJ@q2=eq1p$;a6Q#ukrz%H6VJH%!e@1A0-_hJjr(-^%@pT_W(dtF-(80o7zy7IA}oJ zZ<#(Sfk(x8?$6<8h}0M1U}DEFA3lP;K#B&!S=>J)=JW&1&9|UNJ!M=6Ia^CE)fF43 z0q(f&k&^w-gPIxtuR(s(U8#coMuG5ah%naYo;zw zBQhaIjEmeu33+!z98JVwk0WFN8B{Q>+ezHR`|{d|taXAG&) zl(TNiD`H=6H8D+R^hK|#beA5_s=0C?aw0qYF*3Q@*sYN5nNk)TE`I6m^ucAGaTuHo z!4ncB!iq2!%p#4niEiS>u1Oxnb-&=h>^D zV>M!0H=|8f_NPp(50!0%c&sYnP$6<^V3MvBOz+)ztKw3-W~KAIHCEy4Wh4b)6&bv9=>&L z_*w^)-sd#b6Ez~~I~+GJJ;}&qgVm>7jUD5id3TZYMlW)IhJ%9s7@So;cwZAG&sHzRAnt3acR2WK zB}eO1@h}FG%b&wz{bJ$CIl5PjMnf;FiQUMwl4O6c#=b)i56_*abFC-++l3^A$vPus zhYamvjmdh;L@Ly2Xscq-{em)=+@Wbdfoi`m3h?o=K_u&yFFAs|al>`tW?J0`BW8$X zj=*l^gDulsH7~g|d6Cw>esPz6R^)Wk6L)+4Q!yMbf6mz=^y$1RG4<^nU@3uBpM-~S z8*Up$`8|~Jo*ToB1#cD}vo92g^2GWD)c6&w_U~;Ia*`{=4}PX7=rdfe^tuwpCQcwzkK;&szI zA|;%%y2v&_NN6}{Yt$yo)rz>FFcv;T)YL+JPQ8uS*Zt%D@~1J&g%RLe$}P^PVov|cGsEu!VeNrpi8=ukOIF}s~s?>pv)H;)>zk3XIF0oylM zk3DmM?uLXwz6<_SVP9h^HBeAeKA2rx2k+8oxEjg_AO@l}2%jqG3!r|oD)d=(+07`u zqR$RTb9k4$5Vh~3=GhUxOCDLp+s&Mp8#Q_~PssZ&*cFV&Gtv(IaEpQ;S=C!Tb;&WHJYQlEm=b9D1amosE&#%F?FBdlC_5&IJj+eZ_o)*jgu)Rqz*OX5+RNy_FxFC`fF+)A@#B?x)17&A6*RS6qDtY4g(!6{> ztEZPyptDRw?2|*PBQoQqI$nH@Ri062Z7awsBP56#2vQs9^35>0AVUpkHu9!#BK8pt zWkmg3r57{_KTOORDRzRYC$w-@&uqx7x=AuRm%Ht5un)oj302 zC$DGBSdCHnM_c}lL{mHe}8Qpubc=KBUN&QLaoQ>!5--zu#WEoA(h6AE1hO0sDlPiq?0@(2kEY zyY@&XNUe=|8;n&k45VnPRY5OUUp&J242FmLyyZjxER*Up80G^KNso|IC;L!L!ZOLF z9pLR;{Nw#&+T)Lr&*+nmob0WYKAM`()*^s~pHl269a3}{royGJEf=z8&l^MX(65P2 z{_K!+S=o8+JJ-d5MYA6(nDWI7n>D1KA#WtY6#en??7Bwe7^X{f00BRYhB#_tc#i&v z3`N;36NJ-7pl=xmh4VOM#-lSB3LcHio=Y`=v?%@MB( z5T5be=d|39usvR63rW91Qe&PZ@$RcJVXmv=^G#~ls;M{HH)Pm=Ed$sz4u*7o%0fx_ zo6vHRf>_DX zL1_w27J#UPt4SniiHF`73pc7Dp?8vGrlNFR+ zf(Ye^*dY74XXt>6Ggz-yN|5J=0!RnF$5_&>?==DtcSQqej+ zG@Pm4tdhwVB@`GYukZh@;~ts*U|6X>B`yf1nTPmX-~FJHV_o`_Zd(mX#`CQoAeUz^ zKyOMONWAR#SJ2xrKgo*HCsvX|+)lXdWBUh=G{2WbFo4zI;<^Xt$Q_mWc4djH-32^3 z+4@KZ6P5uXOCI11P;;FWJ!TA?1u#7Smj`hn@Nh=J{2`YCp#3U|Bmf9KqXr8{ zbiwO?z0kT?5@om^HA4j?oI)nlO`Hx20?-%(ud%hYq(By^hoNvW?()YGi)U&iLA5Oj z2i3lWTsHEg28_44fisd+#0Ney_{FHIon78=jp*7C;_JT)K%KVuU!FU{hz>jN19Ak9-yv5*cCVgD{{JA)p#)vy8v_O#F61qKUVfodJuDQxU%jOC2m_KhXU`B z@%h0vk^)YebuE|{!?^MIR-OZ=&Yf5alQ^2r{FidQ#{25SDO;gI=KdJP;daJI99YQ8 zDqgSk-q1RyfzzQgzWILnZUu*1ZGJ2BTiS{vS33)TZ7MH6PHkunXGOgk=(hQ~Wp9Z* zw=njz>Goc}^29x0nHH#4L76fP6)P6!bmhg``##;3)K|@4d}m(N$i*;KGrP_c%hlx0 zcy`tQ(RVX4pnMHZgznu>SpqzhdgH|jwDol4JKSWibwbA*4ZkkdwoHE|sX>r;XnJ{J zI-AFO8n(h4FA8tf%%vTVqiom{C7xx(QZT=WY{$#Um@Z0{tR;jTM+MEbg%Ka4Nx&Yt0(V!1C!jL3#{1t=}-)z zrl3zl1zoR6THpmP2E4({g0%wVdnzKmig)CYI2h&1sQ|&4;FK5_hu>r(i{RAj91$HH z_8K`FC`dd$5Uw#1HUr*|g0*4fPwUG)IV2q9FF+uv&-Ltfwv1QIN!cj${bbcrH4-%X zT*lSAuGfVwfg(*$T#sU6d@m|`(1T5VgO5L&y{79#Bky})2e_IWD}bgu7{c)S{k2|K zYAsQHm}E2T=FV|4rg7RdhTKzUd)Gm9f6H6@(_9)i$3gs~87J>?V&1eUXkVaohZNn? z!EesDx@)URbu00=g6x+NSTysdxKOfOKahKA^`CTDMe9%>L-{PrVF0pr+1-BC*YPx6 zrW!XU##oxkWM%U6aejvcOXxIo4JDIScR0k`T*=q4w`rv~+Ww{$byX81L9zN+Nt)dE znUiirBHX1Sw8hZ8JEtXU##EF!W0n-SKBQji2UMls2j|!+i#tgwLk9 zBnfX?si>5+m4x3xI=|y--UNdgP{FR@a-wckahEd(?uk#`yuRt2?<+h6P`?@3 zx2=uARE-O;ABYz!Xp-$IS5z3xu&VhLeJt@0y2Nf__xcrO ztv+ILVSL?C1>v5ZUb^U_#OhsYn&=z4CqP}KJGP=7OVBMTFEO8WYZCq0ES< zr89pkv~PHQ?Wi$7=Bo=GaB%hH^)T(}*{-A|krZNC0@fs67<8ABy>i$lo^;2dk~ty;NxN98QbpXxAv3QzY-8(y8$#* zG6zT$=|F{8h*Sf2eX4&v^$Q?#;K;*px6_6iXoyyLYx@?a1tPL@62fg#*k7!q=(%?{ zDxKD@p$^UbpDiYiL?z4VBF!Ty0Cb=PU?r=N>+5|i#Sa%cc=C{hX-UX%JneQ*BpXTlF1~nsl3^G|QgrTJ5a{?czt6Nh zWMG$iu|Ca0%NLpK>w8yfVSzVS4~(g-g9#GRu~5UqJI&kIz60oH1SJ*RX%BEya2g;w#3vD>IW&TH$GyNEa~x_QgrsXQtk88H5$ z)~H!MeUi-=!VPd<n8|$n|~cHBOJ2TBrk{n`=<(GofN2}|Y-%fz}e$&aA zjK;b&xEZbW@0%w)e6}|1iiDTpo0e%Hz_kDAHOM0FhO-f-UAyb$!Be5A8f30=G>nX5m)O$UG{pVc=F$SJTR{w9J7wEckc7tTE39Rr8bN(5L(cL2YQiKKGS2>d~bKexA0Z^6|dV2^A}_ z4^wy@7N-uxz|M9(KMvT+9QVFqoP%G)P8srcDy* z39vJX-P%D6Q!g({f>Vm6zBRVbTGqxk`p(2rI^~=ZV~%P1Mpsu=0T{UCMu>FNaK9V3 z&F)f-;iqos{aO++zw*{OxlY;j(}{-%XM;TrhYw=rga@1OTHk$K3s>zrwLWzVj0@h| zed40b>f8cA6H3REMU(x-siP5Na88=K4*Ks2RLKYhwm-S-k7 z0`B!KV3|d`RiHW~$%9b5|I$k2#%%gvWx=0@=DqBrN5`0`Cn&M7PsmaPo~1S292~TynxjNp>})NiHTv> zT3XvR#~(WhQe1(S42j?Z^s+abNhFnYGm8B=fp7dc+lmnz zhP)Al9lqoT5`qptpAiJgDUiIzy$))M1K7X1&#GdPI%o%Z9+hDtr^hj(VVq68$I3Sv zp^-WD6q-=*T}c@MEWT!5lzMh`{ST`T`t#_Wt?=xrseR(rmLdii_i}1`fkFGmIv=4T zoCFybJ_^^E#LE{@AHI$f>Bq{BlVoScn$mc+)WPIrTf1f-?s;kZi$9Wc^t_u1T$LLsB~{4SC*4{J+zqxq zou!lO;)+)Kr^F4x`+ZpTTam{ox}oh3-k-qk;y@LOka2PBq|YHMQUbg)W1sxdN4d~W z5Ji5W)pg20HR8CyZx5FWGGALc%eQ~BiHP{luAa#3`@8Ps_2Ks2ud?%UnIJT)(6Gkm_#-4>!8>9Ct#z9`=JO?x@X zv~Ldsv7Y`>sG}6-YWCrJ{L5aDJFoV#tAkFE412iSkFkvK>4){0@F?iHcXZTmr9Ka^ z77d;xc(+Yo#v(<30`p)5Gas=#+A);HUmU;%_=+_y2!`$-drTn|!n@z9(>Vcp5d+EL zCz|pZb7$16h%zS8FI7j&qnISDi2??G=+yh5_$vuc#n8X~dPv$nI`KwrV@roy{owYO zR~4b(4_Qyo5*Q5EL4NH+zJAJN`DgHepNCEyC}eCz7aQzgiD8NQDk|EC^c~(b4Rk9p zThCUJ#1KdGFmIVK$Lse)igag&ml4R}XXWZ}AYg;@mU+*4(Q6ioLx>C22P0nGFfb@> z-N5#Q0f@-?Ty^eAn$OTniNeFj-z92y%_@UJvWWgV879T$p^S6UMs&(EhUk1eWYOe(Xgpa6#aci{_|fgjvg#=UL7jC z#yBkA15pkmXANu8u%$E|vAno}h-71=Ult)2ORPbf*8Pc>tf5ZUvJ}ze5+f=(Z7)I- za?dlKbzqYUuOUX>^A`;Jfs$5UiF%0gh@3}u|JBCg4vIhjAd0`MwC4rdrNKM_3P<^V+*Z|~K4c-ITL&gQ@R`?&j+?@EuXr2t?fvJlv(1O4x^ z5y3k#@{vCrc^0CXp+Hnw_1&U8|J>SD$g^arwe+j5DyMdsEKt1g%1-+_xztGE@XCMT z909BQ;OfZ#P%6a5K__Yr(K*lSEy2V5nSodq6a32TVYsC#f zr~U1luO&$55HXg~9Xb4B2#q+=>umwHFOiPjQk|zzX3->Cps;X^gBdGQO8TE*1_NcU&`Q%o%Jd3zORmLkzrwPRZ;)$LrGFY zMzQ+pj%i30%~`IfocVgu55HSX0z2aUoQp`fnZTMzMSZf6>?X?bb{imm9+i2v_UFgS z7L$-VrX%%pr)XAW;eM5#BOJt$a~2<(MVTSxAG2CaejfDw89#*PTHpZ85t-X}Hfcp! z5=q}n?INu5q`WP<@+{SlD7M1~pW+9}QaV5oJuXGipu*gr(~}dU}}^1JWtA=6l(}sCrI6qaPdlIhy9!vz7DpO_-(}H=*&;>}CPU3K z!Ab&B(;hX=fhQ>8PGvV;5E*q7|A|&t=SVr_xBnjP9M(GA z{&&^(Im~^K6mWHY71^?Qok&@LTtaHt{FU1ct8(G_8Au2-8T4ihSoL^j?gj+`60zph zB_nR8qNjSO_;r}lXru53FqfqAhk``1^G%u6bR+z4E1UT{-tph-bjP?qCkk+Ys{{@9 z9Es#*g!Z$j)~uuL7jrkr(K=B&5)VElZr=yrzk*zDC!^g$Wri2y&tR2&HyI zjd;rGosr!{HfVw^cQ1;#XT(a5DeNaxY)v^Fy>dYuWL8V+rt(j0xDSZm=HXz^NkwhH z>&vgGAf^vCd&E22M`b>M53Z8ja1XrrlN~80u?}7>@t$cvV9-u==w)tH0pOBj!CN*j zoH=)9MMFsPypOXPa;LoY`rRcwr@r(#g9PJvL;(LA$mN7OSfI=Ebb*?$VEemCG8(B@ zbaraH>bESP;S~M$T`6)TUb0ml|5;d&Zd2MVLyZK3eQwjR#OCIHyg z0E6UUT?YN7?*HGz6m)`EHBctjE-d5(l!^Y;ALO0@z!I!C0YD5Bc!dgZjQm^k`2Y5z zP9+%j0LBk2f%=Jj#rO-nferv7MABL7__P5(X7&rAl;!J#+XzUuq?*A+-v%Ka}x z=B4rJy>%+@_MKAQ$?_w9W1@I!?1V_fZwmiUXR~>D`a~PH;YT#(6FUR)?k<9>{R(tWE$Rm*?H7Oe+pay`NFL|J* zJO$>pdK(?0xyj7jVouU|k1rG|HW5wZTkhYt^0G*h5&>)QT4LMa|5O*?Xt6{?M-fMc zP*sVp%~GHND~P=Rp*m(R&tw^K%xHj(%(k<%`sW5-zlhwpA>&OM2|ayhBqtcntxh(RJpIq z$_c)yx!={$rp9HZmMu|UCUnVO(8=k1`LEEAhWtlZ*-8E54?HEP?x(K#4_KByzcpuX z0KZZ^uS93m>8#c!qOMj|_8#?7hdt?%V!)B{UdXzi@kVZ%=ST0L^w1Oj@Xq%J^ep_( zn_lcWuvK&Fo5@tb%jU-;+Usp8QhYBcU7KB#Ec;&OdRP(OxvCoCL z2jz5-hSj}s4E~GH-ac35a$lvZy4*Lf@^Hxg!EI+ay?zO$5Wv&WC%75A1 z`EX>Q`j(sNYpB8rR4OL1-#d0&Nx==%kDrz`CS8D~pzv@#9CxCp<455T z0MHgARAZ)i?rT^Q$2+gox13k;m!v1*p`0`X(T|*>(+%ko^_t&uhap>YACn61PHp;^ zp?@2oC#fp-kq>9W4(!SVRLZo1<_GssU+H-JYPk}|g!`TKEpR2$;LGOKW|*>sPR*ISUwNZ~A9)b4N0*{Mr(4I?gYK(XA-eyk84XRg4Uc7Xk~+kjJUF zT-eH<67LV_c`Y?t3Vz1M7W~vV^tm4*5(i2O@9eEOPtQxkHfV)diCr)zf%roDb9w}IPZ%V#XM9`mFdwa6V6Jag5# z)8y;D=N&-*jN=1bX_^ z0Nkp7l~UOHWBp6jlKZp78-=BDGk4s7x5+AAXL++9smu~6{bu-$tutV%#rzV~De*^! z2|@NnwM!Zq6HWTNw9oqMMi;Ng4W1Aj2@SZbGlzY>B_aL29eWD`y}?1Yvpjd~+1ooI zYCpD`%*XU8u!83v3LdUxvWg~sbLOk;qDYag=*b~(`KBO_?4f@<{wu>Hf=*u4K~Wo9 z3Ql(P1}Zy!}eAO0gKWn;w{J!=piq{fA@L$kn_J~x*{7YR3FW> zSO9mNKU5s9@CgllV|{4Mqt{yWEj@P^4$T^9SB$M*l{m0^alfR0+C8-Hb7}#*eypLc z4+;0DWfp{;Vx*#*x|VAoo-e65q|7%3)S;g5u=K~tN zt;AkP8Tj7Hm`OP_shkQb{|5ci^#`z6j={DL7mQbpE1Dry9p-jn_1Jm7_it8LDO7OL z-LR=j3A5t$D&?IUGR8wZQzODU?ZpoJ-`;*R$390rzqqqu(WDbHwS$#?qp0C-Aci#= zI(2u$r*)8&j=}TJOg(Ba30I6+{mDGzQIthDq2a?;j5S-D=#aG4&)9R5ZjOa9!jB0J zwtDSrL;m}b9bZ!}DV@seVn@MH(!rUpP7<(?RC1xg5}q=T2l`9QyhnZ>_2HAMa;bE1 zmbj(J@gLv;Q~W(S5`fFb^V+Z2U62^Z58;*u#s7rS@>Hy->WsVT76s3NwC_=2V^cM2 z=*v7iZ0V8L7>7gktH=x!3jhdtI}oihw|eB@IbueX_xrnZ9;8Xx{IQ?PL^(ZZe8f~l z6i|>G>LN;2Asx&=k+&JHW0A4@CVkH_QA18a_r9aDl{|;^P*N*yzm|$^`!jQ(^KePE zW$J;B-M#lAk$1#<7Oy&OcVG`^+ven5oOTa)KhT>Mlu4Mo_m=HIPoM$Gx-v|LVOC@D zclsPevu|t1Z@41bi(*sXG`)!$!B36|Ia`T^?T68bfd2hoCb12(H1;x51%BmTd$@S0 zU;GjQ=NspzEpp(346?^A@m_-3>5WHrs7r z6%8AHMAZW{-HsL(XI-xlHjkM4*RYODcO7HX-S_R4V}wyVHu|kj=;oNW=?m%cMRbDQlBOSUbk=Qs+QN4aBtysX`Og55|XuznvVbFw;A- z_+8u7j{bPI=OS+uAI+-C%kDdRF9%nG_fIq@KjE&^Ovh5sey6WW0ln3|#^sdn#E9k0VJ z{me}%O}SC?gXo`1I8=R%NdbUV!O-#+oX<)qVs`nlVf2|ApIpT%_x-yrhxY8xVSl?z zH&E4xg1MU}m+PCKg`wR~mtr4r$tj^ZhNjmbkK@@tE7ieF4=glr23;mD@*l8sYchZb96j|^ zqW>8ErLZ^Qoqm;C5(!J3--FaMe;uP~%_UO+#F(dg_@e4!Q@PfSJuR7Ak0$0X8vHoP zz{yPqD%-VYWQ~MW37kXsn9zGHZ&6m3KMr{!dXg}chVhX11eIAcI*-5q^Ch#3jUos8 zl$`cxqKbI@Zm+NP=;gQnhq3i1sYEy!;=ZGqFf=~tfp6&)@V_%s7mgg zJ2`D@h0HJ-B!Iv0YDs0$Rz1{RJYjPKztGTm44ZfwgT0y4yv}lo!F|3Toq^+am1Mf3 zR_R|6D0>!p>?dhkown#4cl8Q+9Y!p#ER!%)dwH(w#E+uuRW++Q99ynz!tH-bCq-DE zg-sd+k0z)S=?Ef<4&(w(sGoI?<|Zz|&qW4=a(*1vJAM|O63G+Os9fo&TbXTqVsZ!K zPt=#vlo!*b{2C<4jFY}--DD8a{n`sV+Nbj@W}Oo;+#L-GeC4t3;%eCCQ(sk%Ru4ve zGY(teX<`%l(x|B1&rW2cSNY-RmSqC_@Iv}81eY%zU2}&j5W>Y3$}X0w#0Gi6d2?mh zRXWTY^R%3~3Wl>|+COx!;#f#wFq)9cwnq+{_tK0aqG=v9(q?ORa^leUbSg)TPCqfd z<>kbY8h|cwh)MIzr+m=fbtQp8NF0ou5PdP4XT#tpmYd_QWe241=X(hp?LO01QOj~y z^D9@~o!JD2!sa~C+)mTR*l1?yw&W|W%@6GxIA0gOFKVTxWd_Gw#SmR41>KnMLgy}29DVBN0ZZci4(I?k+5gC zS~}hP1vDM@BI{st^%yx3r;&xc>372TnzzN>w2yLftHTLb2Oy_z64SYIm~*bg6hFN@WO-HthDV8qqH$fp)| zF-1)_#GEInuPGgP?c&1@dq7VW>)R`ycIraf&+<5v=gPzJGlc0FG$YKUTA)!r^+H2? zEvf|;++e_naWY@IAP@6qFR&qhW<+DsF4(pCqV~(DU}@tD zaw59Y?U688;vFJ^b9V{d`q%bT`990cqfQfq2lKm)77<^f6*X!kNCiL(`V8$p@X==! z!m+0!ZxarzwSte9X7duTk+=RDcu7ZFbusKVERwM=%HxgQ@Us&z?>8^OCokRX=icv4 zoM-I%^dmk^9H+*)Y|>9$q`RmeJ+vass8N=@oe{Yi`y?)T#F8p56Wdw-RL-5}g{AjQ zj{b~!YW-oZ-?s*O&5?Wpp7-eZJOU#q za1S6AvGgzL^kV4VFGii8H%x`s^0dI06yfBX zmG1Pqb=wkgIS^*G&y;RwTtV%5e>%=E4wm-X}Y;2X6K>9 z6>6)qZ7Qhe?19izugGee!dNzSaTE4Cj34?i#cR|V+bTlggJ7p+T)55giMs=-YI0C5 zJ5fhKzOiG1|9rP!f#)m&ndw=K&ku{~SySjq7^bBpz)SS>QC_+j(9m|Tl%AU^#?nhaHZ4M-YeB^bUNFO*CpaQy>9Lt z3g<%xLzz-3WT8DeaR{~Th6V30swyaMGIVale*9HzsVAXDj{E+cX777s|4fv8hY%T79i02x zQschQM^4W$x~Bnc=4|Uh-B%sRr5b*~OIDFPMk6D2pNP!{}w_EKyk z@N@^@E*6H&p_#4gq=er7iaNz1R<6^E>(2SB1IA4LesrE*^HcZS*&-s(cBRCc)THuewi~a*=00scJ}i{*H}{2rpVZK zgH{P`U66qGHcPgMu?rPP^Akuoz6YQ0aLvE%aZ*+HrTS(I^VaNp%0}!P zzQk(`sRtHmQ}QMs>4`qePXfa_`N^cvRQXa)GPIHj_=UN)G|%;aK37*x)hi< z!Du81J~u^oICx`4fbt^y!zD051k1c@aodm>DcqFSV_4}mPYLH7ttzBWqD}+j!sIXO z{BbVIVYoEYu=QDk%c!pHii9+Xc_X-bcAvzy53w#Uwgnegh|<$9`v-7arrFi=N#CY# zId3`(Lk+&qzNU2uN)cs>8`-j6#LRw_8w$4o4AdHw;>}*~?>7 zn0lphpMCb{{>hXc9=d}uXF8o(R#Sef+0yN|sLaiemSi(%am3igrJ+G56A7(B5qrqj zXVSPn*pxSDa+UY1pS!^QFTHwz_0jiSvzI_A&2`rj1(=gW9bMvLEcCv04ofAno)2Ca znqZzSDRGi5q*rZkT5A>>Lbuj?3zO3lQr(A`r-xS%#stx5b^XIuU9UA>-8{3+k2}a@ zv`_NFRxzMuGqnexXk`boejdzddS&-oFXtH3=10Vq#1_Z%r?P`cylI7$Q|>Y+hw-R= zwciiRNZPd7L`?J5cRI<$cbs94akSHIF@jDjJGypK&3(cNS7(&e9Zn0grM4D160s{g z@G*>f$&pTbij(t(@A0uW*tLbyR1f|bF?H}Dx^LJ?1qKSXTJj}@{Fq>tp7#nr( z+;IUnf{BRRu%KwE<9kN{-Sjaso_jnTpCa$b{_M2PLrB{^Cu%2iPs#_YORB}5+FL)| z*gTCDg)(UtMXkml*%x4{@A6tO8eUz& z$mda^sa7_=b(Vfku={0KYy9kBI^k-@JA)1ri*_fxL?Md^c^oPH%on3x_EG2-Wnbj2 zc^O)y#qV)Hw}Q+4ILY?2wo>O6qhFKi=mzE5``>CVoitzcxU+z+W^?#n$#P6q61xv+ zO;T*JuAuF};tP-)bh-PbuW39Z0J>}ZdPRu7#?RZ&6F>Cb)p3>I?k2r2uN0k%7KrhH z`SyqeH7{RYbEK&j&ddB+zcwJ^I%+1~C~T{cY$y$nN^^FNZh45~Zt9veZ)m%~5=?0{=y{SKRHL$1vF>~zGwCF7TGPD(AEt>A>#Y*bM!2e2Z z7cCzp8~AJhnfu_Qvci3Gn>c2Hmp?Hw+(#Ery6g*7Ek9=qhlfF=UQ2KL)eDH1uIHia z5gy%-l+_r&l+K#zobV#$b?V7g;Jk19OZ0wcbJyDqemEJ_ZX` zSgc~0>z>=T4`Wy`OqKi(KUgO@>Bxn~yOz?b|N4`0K(++QvlwT^$PTV)JXc!xc*bU`H1+Gkt#-U9EH2^c)6UrM%vv*RYwpkDVK z$Wi^b(BR))4AS2TETXAa#;}jQL2}8L{uvs~YzZ4hiAz1PUh}~};*o0=$DL~(p2%6! z^J!nak#bSuf9I^r{m}4!$(JjCzb0v4a-KhsB2toxZf&5E_9)z{;A~j-2_M+e&-ZVC zeD&Y#%Ypi<`UfaS7Ox1aU79!l;o<$RJfZk=^I@C+i1~Mp#I7lF+P?=cb1g3Px82LN>{n;WH~eH#72c z^1DLeXjC8=Eb~Ig-ct~bf{-!R|E1;5JLzq4+P%O7t!e?i;hRp=E_1=-M zdrH@KHxAOa>eZEl4xELT9d09|3bHVBUXd+>bgf!YVAp#Z`qHq`Gh+hAY8X#XR_q{; z49ep~>;)CTYWE&TEY7$M27`WZ8ycuuvQ&(2UFPfy5q?k=Z5zY5_zS1R8-8Ml&i z)dR`shgo8o97>k^7>2Vjx68Y#X9nx6qeIyj1VGZKAjKyiMw$jLMA;aT_De*H0T&Gk zxt|4(He`ila;T<&tOBj+K!S+1Qn{>G0icw!7dVhpH7@Z=>_3f0kqTPu=!aa4R~+g+ zZ3CP6SLKjE+v}SqKVygp#vbU4vOIjhb1`faN=IB2jP+FonKt(;%*RoT;sCchZP6MP zDQvA)0GuC3_c=cqGFZX2jfBgev;+!i97E24tZt$mIWeQaohW6WK~{W1X0OI;uYDqs z^8SMJJ6Ic~{T{#-vkAO%zy%RRuw!=l7i_Fvq;?}l-r6ofV8P}TW={l)-OA`$G1(d#hT}M;3!*$s$`Fb^f zlJPb-PuBGQqdZ3d0)T8Olwv!56aoc{e9!*+nLb7d`d0lnnkx59Wl3Y$<~v%(atrj@T5GEEs;J}~j)NW%N~+%MeV%-?->jbg*`vK+=}CBZVQSo*%byXu zX&xlJ&FBKH8wl1Y8wJX<^tCia;-F+FA8j;2L2y~+HFm0fEt8c!pYUY;k=*Sf>0LWv zE3>zjUqyOlf6JJdJ4(4r(zAIQaKju5m+~Edek4xJF`l2I*;&0r{Omjq4WASterh4x zoWTX9d(#>OmK{^)PRh9XOx&);wa18T$a5P&SADSWDssnv{+UJ1I;ZX{pe=I8nfqvl zFO9B?8U5jES}1C0jIKx4=L*}(D<2m*Rw&v&nb>Cfoh%Z(#lg!w`k7iTF890*z>fq& zY5*-UUmWB}tC{aMEE8b5J4w2!$%(H2Ba<)oi!)?3i0WkXzlbiu#v&na=0OPm{XaRA z6V(O4(i`cpsr`(h5G4Ia*+672Ccmj6Wp@8`-8~)p1_n?)uK)C3Y6vqg9e+_oCXs&D z_%_PYc~l=ct~`i!#t0T#s-#VRxe(4zd<|C336kviYQPisx+1n#bO5xc+s}hX4NAS8 z26i5D-Tz0O!!oF{C(Sa+t(6V+&FK}WUfUcA?qN;(adBiOh~Ja`r>}!bz@W6y;oqo- z%@uEHQym{+t-;gi>wl&_eVl4jhDCytY<<^~@6}mWa9D5(zjJ4(fLS@2Nvb_;p5<5C z9Aqc{)V#|7D)f~ZCcZZ@2*nLIH{9Rmu*y479al>-2%m*SM3K?mAKYX>KRcyPAA^_F z$>VAr*cevNkFA&(joGENe}sw0ofZSwun4^f_# z=-6+@i_SG?_C!Z_tdUDF2w={^M=71 zjvtaQl+QdApfkGvnwatqu$YnC8w)Vz>7BNy8Iq{J5R*GYz!Cr!&>PS55LJb)Gq`cV)ZEgd!{S4{wJB=EyW zdi;$4pFb8!?4_Y%9}*D44TfEVg!D{v?uAD~Pu=~qW%~;^z;7Qo4{e7)w`lA;-pGkF z8eFep6M=xAUHbj&$ec(ArOHA8=LBWoSAt$jXF#drOkzmhnt2B3fB^%&Xl4~y9s-|& zz(MJtDvM<3B z6b;v{d|oSqy43lwi$HPuf2`O;mnt8*4vk-NE5M7MUI)=4NUmOGp$F^8eLp!1t=xvK zl#kdp=1(2?brTT~fyLBT2_zYsAogIfAyBXVZJ|G|0tsEE=J(8otYJe^(Bvi5pQ5w= zXAdBO`roH6cW#2O*79CjYA>Eq4RU!>R49c_uXQb!KFzc4P*PEs?vYM2&tKq9juYTg z5dq&eFd)1FS?ToSFB?4`SW+d{JC9j_nZ+mzxJT)0eX%UN_g_Do+6%bU39BRj0=v5Rw@_l;n;OgRKq;*rP6cMspL?iVRq$fAq5 zj>XFJ1YDyqfXh01c(eQb-gYhekw&w54mw7#d~VHlBiNNC;>1`~v2I$BVRm0iCUMy( zo`MUP9{uZMe(2jMAXvF1F11%1mw|MmHGx^ezPX=*o`scmg)>{FfjspXJ5hEUS2IUO z4LiuCVviplVB4M@Z%F$6BErvx!+Dfl<)9f}sw>tWhNy@i=)xwPMWS9BZ9^*rKh-bh zkA)mAnoLnsaM0j9nolwVS3Tdg6Em_?^!6+CMgrB+i+kVVGVadT&{wdw%DUCP%^7M^ z#;_R3r8z2OjLuyvCF@HiE+a$#EW)%r*@gy&RlhnO`Pe5t(j4#y^}CQx$^TtyJyJ&S zhjXd_O&x$beRHs-nc(^3s^fJOe1Rc&6Nwm)e8M%%(*rSJGPl~wQnTQIo8 zbpdtgJv-9O;aSIc-0GZYtE$p}!R#3X)T0Rh2AjfKPUw^`c9FUGq|0keZ7~naXB2iN z*^T*sXD>(g?1(8-3~q!plG0|j5D5MTQo)QGRXB`>(ql+jmJ@VzPCjl7CK2r|qgl3J zWMc08)b5 z`Wb*Vr35Nrk ztUihjyJ(ClI-*k2Qrq@nd&QSZ$CTsgi)ueuhS(ABNbZM=T(R>R=~ zk{U({04^r}FW};*P!TuORJop!h*iJ_9mV>!=l9Cy=X)tY9vxf7(-1f@$?5)MCB?z+1e5gtV;;!e#3vBaTC(pg zJW`~HSby_ZN9lwa8J#h!nQ?m@qb2oILFKbS7e4q-|0F3uc-p*38kK$j+^lsI%Wq&fU=o`S@*v|HMRAw`LBq z-EB8BVT-F^f%-jL6@<%HR)x+Kq@oZG3SUZ&yXQ#Bl~%7uPSWXyxIT|^=?zw@rLEoh zUciVpmeiZgA#P?*Eo4u-|G_e29nq$zI#Q>9rIjYnz{CumU?3rjBsnyY_P8F@o6p-` zkInb|Kx;6vY;)j-N6tehrZM*H5o`b+Rr*H6JTXX7=k|_-X&f9g;QK<3{tAaaVnY^H z!hA8|!kc`$?_IYm=aUL^aI@1rA_Z^5F(_L%s@A}r788IPn^#e->UoNxn?d)k{}rrm z&{j8ZpXtxKi&4i-wubS$eF%~J~!0VzR!(@la);X+bDVDx8bXz zf`)nb7H)JWh<>QKvEZQDC#NS)5q=h#Pw^cW_`?(zq@>f(&1L@G`0|xe?wGkYGjWWD zr-ne7hGC~j!N{7BuEUOsD2LWty-y7k^;<=kIS`bW%sWc>5W)0B?EH3R@dWFWjYvV- zv=OGbRL_aF6u}@(k(gQe+2F&c85_MZ@I2sCh~Z(xehbt{5d^~|k%CZ2s#FcW^sM~^ zkuOk&8j(=2uzw#mIk0b)RN>GgjJ@7#BSs+;oFVaRp#CWbp6`7U{_yWKN4 zI0&xiQQtAiP7V2!2G=aYG>(HAc~~ zlEq#-XZpqXI$M&Ucjbr*N(6#mfOYa+^$N@I>d2ihZ1}U4ENx~I9y78gMHwU6PL)Cl zk&<>!zP$YYWe%Wu2fAM}IH?1h&KcJn@Z6D3eq7!WlMKKqkhAI&z~!lF^n?JWehCj# z7b|SfC6Q6~!%zy1V|HV!7(8G+y)P za&Di7ssHQf`#n9|=7NGV-Q$Y8v7*j!aNql_gVLSpxVucU6gBt8_dpQa-NE|d zzIV6YT6^U_@J6}t5OuuplZZDpc=W8B(mXxcLj;|d`g@vIN2d3rJ*hwE2y(t{*X_05 z%=E>9FDZ-pNOi%IY!Eo$4jaLth-u;p*4|>zwcLd|MNoI;)NT3gVxOl~Y!c(01NI)n zqa*2tUh{7RP3f3~D&l3|Y!^Dh;{>$xUsZafwk1@FT_bhynYnz0t8Zc-jSEVw-Ral4Y-+5JW&W@)%>}T>P2S9R6`~(3Np-WZ5GTUdXLjk z?=en4&sXF>Fg24Xms8@Q9-l8%z-BPVb@-+=rE~A}ut~k0rC70(42@V`CO);87{84} zoBMjD%DVRC-vI6IioT<-%cgP1vnQ>MY-nbuAOv-Zr*8_j)Amf?SEDdSF1*4SH??J? zsr+$2LM}Www(V44_djk$rjFFpGVn782DPcy;sQ*k7EOj%xb3o&lCO4_HZbe7uwHKd zh$x;+wY+&Y!&a)-R;sZ%V_JQQs;&Wxpq6e^)kX%#E%VA!(Ar$g0AOs{s`!L(24~M? z(ir|ydvQmb)^BItOot^AX0OlfZRgS)x{#n8d%SKviHyxDn!h1%Fu^lGPd%{$a2iFN zRvatkQE_St>FPyF+CcRWY6%#ry^pXeg7=3?Y-X#ka1a9#))^}1oO2EzdLF=~7p&JI zAP@KdJ|UOXi*ZRbg4dZZ@mX%wnw-Dq@wh(djnhQXk|oY8}o#qn0Ch zHmK;=e(&u}Qzxfvml$#Ro->(9$90s6eciV0xG7fyM;YLqezwfO!@7*W?F#v(&Ya{K7*LG;xn)bcn|ydZCBB06Yz`7z%NFnpLeX~HkyL&-VXE9bfwbj&Slr!POy)83{(2tvW_eu-fjcK5Mft42Dv ziml!un8ZWu-;F!^Q^b|A>tD%s8K($c&$3)my_dz?h$GR=i4)a%#xk^my|;kPchIPq z$g1(Fx?y1mzK%gf=pRzH`;?4e0MFJutJ4c`x|K8 z+vQG|h2!YCcHVw5W)*H_*RRY9ApR$Jfw)2PJ%&S1d&s;>OZZxxX}(qkTK- zVwwsr9%fXL9S)9eJSuSF(FRSL&-F2eW>?M$y~fSD3}xrBh**tww>g%k?i*2zXn+?7 zlw$L9Vnv+IE>=Dky~Kgbz0Z)AQ*RuavQq>XCI9pz+n@_&UobMY*MmrOI*6V!aUIR7 zk@10S*R_@bcfWaN>W96=2^guq0}e*Pt&baLDb5sHaml|#w7pE`F0IS2ypUCSg)PY@ zp{E0M%FwEfUJbl0Uv6`y5EmhgW>T^8-^OKO`|_L$K68GbKR^7t|xD)aHmYTfX4U-M)agOnXaf{|I}XHui_D6?2h%pw3q0>w~K{pjvs%;g48535-v|Yo9Ogc z`5wH2ec>yg{q08ApLLcSY|7aFI_?{?$>-#Jr3&acq9+#(J`o4ybHcI3y({+}J(LCe z=7+X}o-f%n2)Mhup?^fJWAz2BBX)|~TsoJqdCQMH7hLvdr`Gt_S3Vj#9}K;=qDMKF zTPnZs^6zw8B>axd_3up%NnI~R_S}|_Lw2l8b#E8li0946E#TRQ=!6#22Ow_ZgjJul zhtcPgiY@Hb{e9|N4kD$Svx(%=YB>bB&Ll7gQP?c8Z8d8rDPOs=;^9D$fmrB^YnpQ~ z;hz}?w7<%`>*uEYm$|-FULtoC2+OFvx)~MqS*P1{VJ}u7PA{_0>AjeyazrCW5hX7- z2^ugq4H{atG|uCAB7{A{1qRhIOM7QT*@pr)MW4*6xw_~D=6incn&V9kLcnjcZP+I0 ztlg9GEazXRrzxi^B%NQ`v4mos*&ciph!TzJ@&?hJKFs^d%pM?GyuQ=W{(bq)k!^sS zHysc3C&!51h~Nbcy~wlHgvByEtc1;CUUzMcrQuKae^U7N5o=;S8XN75;SKV_WvoIK z?R2cjxmf)0Mr%}s8*d+S90{u32%*5wRu{$Y)1JX3)ExJ$IW@1)dk zw|A?qfq>DTCH3>>@b0fL#u3*|=e~J}{?6b)T~{ZR!l)mw{Ip=Q=~ufwQu=vznw+ep zW4EtytugVvv|rlawtbWn4cP6^Dm*&6`S!nQsFU}T!pvG$c0a{k>TQe9%F)}=Z*IRW zPxI_vn!ec%V9VcT9|#L0og3HmTrjRu_jQjH^r~E1b-)wpY>f5#$*y3F?;S9L0_{+W z@rei-r*o;djEUljW-kUyI|{V_eS$E; z=ZCFQwmzE!p!`<{8}jf|8z*6)C}8h&*&mc*WmRM#$3|bq4vb+LE81DWJ`BB&^ZUe| zR6zPM;It&|iwb32Fk}Q-n|sIWa6G_+&QY6$qj;~3Sz2ggDqrjQd#K%*Kz0r;jiQ2& zj^`=ld@nu9N61jzkBYbp@<9-l=iK5Mn@aVx&DpJ_$;}uBH>Tquy4!c>o#Y}WZFS^l!J|8d?3Vtsi7n}oAJ**1 zKl0ibtpw*R+a2tMkZ^J^rN-d>_>1EWjFz)I<)CaFE=~_BW%rKQJoN4-s`o%ih8(k4 zIO+T=V|SAOHc(ndCU=6MNsN4(-ri%6^UnJH+IYn3?NJ_6Ej&GmdCiu5;@VNVbuE_( z)^g5%&$C@%=k=so5;0@E57?2wz_}v%M&K2f;$}_t1xRqAu>u~%ke81AZY(!AvK|U8 z<^g^&Ac;iTEq27(K?)N*6T!2drXbdnyG~+*{S!cXdTO_r#Fd8f=*@p=`?|f8M?S%_ zqgiL|&Nnuz1^lRi@!QID?82l_3ONl4*hWz;AOB_!ful16xIKEg!~nbBAfsYD0X^uE zcZ2prMTRPn0jhA6NZTWt7x(;r=lkXi6kw4ec59jTQVq*5FZTqw5I6mfnl~gy_my=O zcSNp%FTVe8U*st*>=JNt*jr*(tO$U1=)n<`{*tz}lBXgE5AgGc<|f%{A2QmT{LU_m zY@GC-*Qa&Eu1U{q?OO#<{q>J2iG7WBLybd?PJu`hj^BY0e(v9;OK2l_;C@Dl__&X)zC%G*-kEh$7!)wwD;tRhK(+%;kyLU24^^Cs&T8hkCVl5=?BdE30 z`L!8^>1BW%1QQbfw+S7-5_|ZhJM95jkq>G6LkN$8$MxrSq)8N9jHtW-HbeQpZAR)# z#}VFWvxep9%fe?@X#Ro{bDU7&+18X621}L0t3{NUy&1mM+38PgOHZITqUgGUzHWG~ zlxv}5StnS#jT^vbW6G(6^#Z8ET!>@v^)Zydd!M2`$wbuUHzmf z%Y>lkxC+>FaX$uola0o_V(9%Y(zdJVhe~w&JPR}LeP)lJxpV|ch91eo!92~ecWw)j zCr;2pf(R6!ROmLbPCo_nCIkZcGkf;tQ0|T+6MX!l=WJ(BUW0pV?88wWt29G?Q_YC3 z8Pj{S3K*p!$!)U=I1b{MfmiV=OzuY1r0Het=EW*0^r?;nAO7m~$`BIhg67rE49SiY zX{4zdWd%L+%%184`SOM5^8 zChZe`;IybRHlY|{Wf!O2G}bmcnL^wwaZHZ!bIt6o7vmbbW#2cYEw`3+cc?M@Wb;=B zi6GUvE7HrU_X4PWM1+ZV%12(W!RM8|L=Yj1Wry#afq<8MU|WKqW}#4}(}Bln4iW0l z>wL}Xvz=rkQ&T;=7OaKiqFb|vMimJ%RaRya(D^G?Pb>dcFuoP#LU%SiSVTkLZ$}ft z_>7T(!&C2Y;MDp{`hYDyz8Uuw!R58G(Rv#hk@uCYym8yy-rX#)0*2VF`CdM6`?2dGw5#R_G*os1;=Du~-tuvPsX3{0I|#1Bn9|AR$!U>* zOPj?lSeo^;UnnIbuNmIXNKmsF^i0s#8r+z0>vF4lD3}lhnYF7#fWDbeTr$2DcH7&2 zf*>RqK{1+l62`{^bbapY3*uH04f1r{EXVV7(U>Qr+05Ief^Hg-IlIl;+Jndq`c3;o zwW`9vAnOb3C>I$za%+A^q7G4w_dTXyE={`k0NlFE^79V&{F{f6WPSmQZVNH%+I!c} zlU0|jl7!SVQS5`45kvyfB-MQ2=iYsx!7U&o;-lI;nXkn+iQ4=$xNScxmmg?1vWD3W z;Cj364BJ<}q`ARexnzCGvB#R(5%4jAJErgyEZ%e=#beSDKOyCCRUJhoz z7N19amA2~X;YsfcR(DM^11GR3h@RJ=;ZH0((;+ACi8jt+Z=Z|n-zK%@w(J($X+Lm2 zywFBpeI_>;g(ub)S+$?`z8{V4a=Fs8gOda*d2M!TPhJDMB8wz)nuC32!m2#yCHTBx zQuP8u_frQ~%2LD$A8@(7jB<>+(CG|nkg!>6Du@WUc_~lJ_*0BL?C!nZf$aM|!SwkQ zFu&(I#$qkegs}4$+D5q{?w>Wq(*3%Z%5H#4c=LJO;_0r}lg1Mz4Bl3A%96dly%Vkw zkMG-d%xgTmsMh>opz4i}QTK8*?Ae1%gNk|;LC^BsONsQz$V}F(x)?hG*y|I=D?pK2 zH6PAdk6tB$M^s+fIstHPkpE)Yo6bX11z>dhuGODJrYERD(NN4A$9v%j3eyzt=6^s5 zAavQtC|KgnnJq-Ez|XIgr(mzACmHEcw{|FD5A;}!$FJcv==W9VBRs0V`i*O$ zjuFNOwuA0#8!2G+-GbFm@`NBYK%#w4#wWicaPI4oL*>4>{k4%up)W!X)^P%SJP!^& zjxaj~IE1|NB1R<+#-I-dO%t8*bHG%5wUYtBHk$por;PMM$z;T^3qapO9Yg1~hLw|G z&ph%Js8&>UVBTV;<80g?_}M>K0n+sZ%)R$wm#kY26-)(uJl63YFwI}Z3946I-ZwU| z_B8{e(-BSx7=#2wgUaoDbq~8!j877@(P7T_MfSoIUIdS3F;KyRkye`p*~K5VfOp;T znlSEEBp)p{9pzR3Qst-CBiGauZRD7febGELqFwUfW1poE4bql)&f_)0elGW{qEE(mVB|>J zMcycIRSyDSW!Iww&0O}dQ6#hU2Q5!KK@`as?Kv@_dfm4~t>OTNbc7|9_KN+Qh_zr~aDq`~ zd*Csz626x$y&jCGS^nD?k^s>v{>G#|A}Z843$#=ebNZF+ni1*5-O{;RguNx;8M0SX z-~YlAIRj?27c=;AW~ABlkULSHVG~?q)K19xWkKB zj4?@tyiwXykQtyAZ&>DHSXlJ`Lrt9hae=W%)T{utenh9jx8DIY$bEODsE^SXDB!US z24G$i{tv<&QK4R0AzsGJ04N+ldsKe%0YQ{~xU0jSh)Po1hkt;LvwsXIQ;Rn;)5PL` ziKkTJIcC3oKotLne3;b4j}4C;>D_>zviIN{ntLuC#!OB}Wgy}}hz%MzqLkDY{*Q6n z=A0A?#k$9kEJxt%-wTF#oA{DVS?l|_Q&*!%`;~pn_dn{)E67Pf!0Epi{47ex99*(7 zGHV*7e^AEHd)9eBQPiB5#N)B?%>UoyNqm_N9b2v8O~U_C*qYNcp)i}5_U%8Gx>0|x zbOhY9I+evPwgDpUfAINtv{MdlrifhN{1+a=j)&=QLX5%RZo8HJBVz)|XD2p1`adD0 z5Ooe@6Au&O`aAKDf+iU^$z?N&2od7ZnE#!4Ay1R`cZkM+ zjW@dBP&0?~t&By(Vr}q|Tim98+L{wYehg<%{6W~0^K{bR?BQYO|EuCiPw>51jw!51 zdF~i+c?}U9QCJ(|Z~o#BX(H)yPaKxOkNrAg5(B-wn-$`WQkxP(rCv*>wbKgC&j1AxR+z>(#saE-!m;fqrbe)pT`$NkIXNmxx9gtr z?t$Qsv);#8?zP-ml5JYhy&aQCz-)<=o~v6|x=LJM>v# zr1CF-JjTb%r`#dsn|e}cd$0sGjxxRBScUVL$X&42NNA4d%?fC+kwjpehc_z{UMvn~ zFJ+W192N1tCSc^uasU!H>p!&DDK79x>yI?*pF1idPKolBcf>K@uZy%&(` z=sY#y`=z5U_K2L^Oh?ZlP-soU!5x;cy5Tr|T|d%)I(Sz__r?;sCdpv7Hb^mHWz8|& z()F*L<+3>nD9gp&b_G0E8HH>y-iun3I&C5i$=lCt)HX57kpwC0JfmN=>LD()bc+N9 z7|3mdf|c1q6K+GgKPwH&HxnWm?dc3W$d`64wNTlR9rtybYa=Vv=(djDxQsVm_%>n? zMPgCl%8^AzLxC}I-ig0y-ixj`*0zbl3VU;6h#8M|4rHdPOfswq*U!o8QOLvJewUzK z;v#f(3H5J_H7R{Uw!ZjE1(Z_2t6%%_=?cYl=*BT+>^m4cZe}QhL?UaG1x5A4DoGz1 z@Ou4au3x;ooqyd)%GBxf$40#z`;wUizx&FmInXun}6uo;q=rEu%a~pB@;M&o_>#X0#1&QuU-q-om z^evrZCRPB+X#5ye<4W~W4{5(Zc@7z5suPcdN|AoXXcRLthnL^WPupJ#q|#E=ryN1^ zhGi&qCBqHqPteYMndh@PlagLzKAXWz-+*j(@i?2ZvhIeKvug>hT>@G{GWgk~i!TL*0-6=jZ7pcyWVWUujQw;L%B@nb^Beg2J6sYp8K1 zF<$WCNtQJH#jV7qj5wzmr@CsK0WlnfgsetWg+NBFR2_*b$0~u@vE<~oA<%y4XeEt5 zI6CM$aiR(LaS#sU_2!vnru8C4`R@6>A`=qhhPX(cjrQ{jemGv!HF?d#j~D{w7yEDA z{^#rdt3yxY>A`J8@*Ve6pu2b3jlZn{m&$?K!Jw_#%_{G8iqc}zr)fc}`@M#jqr{lD zagW_pWiU5lNHUFSVnIq3p=a4Vx#i~c5bc0A4}lC8FL+z8ZH7d5@MvPf1%BS62eFK; zT0|18b2)+Pwt&!=%GNwpp5|Gv>HyHgSIYAl3w%*A$;o2Kl$2AOoBx^o<YYwHFLGxbW6KtT67J4`R6okAMZlR*=&fEMGg)-$SGg-wu1E9 z?(+sek3339uyQ0+0WEsle@81O~iJo z35$Z%nch5bgp#yUOj(&X5Jc1b~o7X-E7SEpjp3exr{QZ zH^%K1uX|GLX0=Asdw0VJ3pzjIZQ8*cGT5wLj?e3y zQbCG^JD%~LepHzrh7t?T@2)(P$(zu1?Q>sx3^?XXqzR=QZ_y0X`+%7(KRI~2X!?t- zAV^V=O67a7?sA(PiQ#GTInD%7e76(zJM-SVlDa`A~%AyKfsrDCMeT7f% zcI5O*c+~&k;PGu58&~ukXI{{_C{6KwCMf zy^1%v^pAJjD0dfn@k#Z&=TVW}sf9~#@R!{-J;S}cH4{Dwt!T^kK***$nn1MVQjV-O z{EL^Ruw+;hY}E1He3wDPR?6|9U>JG{GQgPIomkUf=hPjjBm|qX zXXb>B&ULP2=vBhpzdE@mn@&;=9GNOSxXU+wPn4bgAg#*LcBS~v7kLyJ-f>#h3R?H` z(bU)}k@4Ci(-=x8votNzq1v&K&H!9`IIU(;bC?5`#u21m!T_qMJXiy;azV{S8Sk@c z5TxE>UHaUnr&m{`ecwuIr{A72)1o`k+yMAYcqcy^#GaJA0-p4oS~V}#(G@8Lqe!eX z=ESu>p6(-sgPyv&i#^WqoI0olnV3MlpbXo!I1~Yva<+o8N7h*hOGB84>HcU~rp+9r z^+81{wu-bdFbIb_ugYWAVGrHug3*3MuZ_AdL`~*x`ZT{Yz@qF99%)(ey)gu10>2I3N@z3uoyIy~ z1D$HVonPG4VC~-a**e{7Fo$gPGiJNbwDk6%o#CG0cw2&b&6ziVwXN(IjBbs2(H7lrf$T?Dj+-BPBZ7h8eGn7X#~2Xi{qw(P#UmY>%}$nbp%0+mOike5(u z8f2+xwL;yD5zuu&aPYwHSj?d7fTnBRNb!e;^oPPN>k!_!5Ro-leKsSm%8i;=g*!8+ zAy3+JRa6{gN5>Klsi)VY2~#2xGS@ae_j$96^<9!iIp@V~KCe-}7IQ?-<7hwinMuvJ znwe9-*x3bb_)72KCHO*=>$bqFp)+fqVxv*0q1OZ3wt;yBH-y<|uQ{LaTM`7edT!11 zbe5;p(^Zc>S2cd+gV+*5%gTzhfc8&KQDlM2RHJDP;7V>^a`T9*Z~knPRc4=N*cjKB zx-U8VWSk^L5g}ZeG1KjQy1VJjn+98_IDrC>Lyg#5#%*15I|2ePIpcb#zs~pN^cci` z!s64@(}VmDY4Y2wan{Lv+l-pMMS)#Knn)4v*$iw2%__BuXS)L0RAUueuT+qX6*OjP zOJIGCsnqi=p}14BVgNt^j=b>mU2YZ*yhT-?7~JLxaZK~xt6z7$OT#hM#@7&z;BvYLx-4`>fcH5=PqNxFPMA^g?b~E~sf*F-d z=%$Zv=6B5C2xAi+1B2?rqBjHsjtd>D`=!cLUnVDc=;&zq6x6D2GvZDS<@_@D(pfgI z@NB!gLtE(KWh)%>-hnwDEkdfH%1V2=GUMZLB&Qr^{o$3ql+>symv?o)<{XT+X_KDa z?iQ=EYt_?189M1p;odE72I*~wtvxz-AqVtbYp>YOI#k-7(8BQHg22Ql`hyQR_(hc{3V!TM<&sQO)QJ8YH1{^94fHXYfWE1 zpGm*V$!6G>F2>4SBf}|T|J^JtdxZV9apkM;`)(MSV3LmGn#+ia?z34ACe5wy>R20QA-51`B0Ko`0hSNTJ zV(O}C2{V1waNZdOCa2al<~D1Jx5YT7k`Ts%(PG3*wQE#NeNLLBK?e4F^+mUC_s1U2?Wh6bc}q*_ZadrRn5+}lT~c9 z>3+f485ncAr7=uz@CdxwCTWC`3{_-d{1Hv!kxuobf8O{r0qAJ^Wrj zvJee`zwiXHQH|G-U&Yi)6D!IZ8e!c$A9AoE>WWK5%&NEP(dx>IOR+Z@uLD#GMXSFB zb?Y?&x}s(P@QAWG5Fm-G%Hs8`2Q7ip#hRT;l*pNw+3gP;rzUfnM>^=tQK%R2gn#5^pv6}qb+q?HxzgIN;BTeDSy)f|M#-K8 z^FeQOs>cvXMvZv@JIZ3o{we~YMz@ahF9Z*8eU@Xt#RuxmbZmLyO4=UfEUz1|(W{F6 zBJ|(&dOy@j2WB*P6o{(P<`~9SiHG+{ik0M1WORQyJaMg2ET^UbsNHIZ0hQMoIqY7K zYl%H&oK)82-rKgD(EaL`A8@YU_x7NoQ{f@VV-qKisGlSJ2XP(0^jCDkPhxs^l#YUn z*2Rd9%$S@q0|p@rHIW-u%YmRJ>7BThUIj0bAi;OGz3p9PI3}k1^?k%K5jh9qVnuq) z?r7sT^p85jjnMfwdWDZ~-8{ZV|l>A*{0}b;~q*LWPd)si^E6)wkwesl5Ds zjs6OXkk1ro$kwPEIL2NAnf5?XkGZ*`fxM0WiLPT8@%_Ui#jl?6+ikUorrU)p>_%y9 zBB81`C?kE0u&0>PwZ#1)vrlmQ_ysWGqD!Nf$CaWzBVJY&g_a6 zfqj>?+qq&+@tbv&Q&LAO+;ozb=X-*CD^6Q;{6utAhqXc+<|lUDRa~ilDLCU{7wq&T z_D4kVM;V%ZE~UDeckyel;FMC^HGOg1ZDUAT;>-H!iXOPu9f%_V_qQJshaA8sb=J0$ z7dLMvC+qT!OrM^gj&#%)>8zgUhP>NI{;QTd<^n0);Z^zBED$6s>c&@>e*_~!SW+iD zgXYh?f$wFKsm2@>Q`e5SQVfwRG-4HX3kw~Cp>yJ1+?Brj&{!AuSrJd=;EPjuPfjfa z0Kn#zkN>D%OuRe-q_IYSv~g%5yxm`A^eP+o?6Gz~(f{ILAIUNq-Dac_cCcQ)4`)7a zWXTKpO<0Bm6zetuVMZ-&<(}#fbJQH#x+(*mbXP#70C&=+pRHAP&+~ex3#i0kRX2iU^u`V&!c5`4w;Q z2_cU*z1YT~$85RLrl_ghyFm943dZ1v{qU-355|=uNv)&HMMh5E!G}16Q>I6=DIFe#6>tJS)t)`%BKB0(%nEa}KXlAZ zpQEAP|)Ns#DCm`)^)PR)ogYOTeDSw)+$ zRUdv>#OaWHdm;(dSiR%WN)$kh6O~c5-qj?3t;dMfa>u{&Mnow-?iT)k6n8u8ES*$G zbM43M7IY?96RwRc<>INU9i7A{9y+_LpWrxNWdvR9mW1>tp23WmAANjhLsJQg8UTX{ zd1PEoNsAq)K>RI!{9jJ}D~F5g;;lQJ3ye?fQ5Y?g7|CUh53q?JF8aKUfS+N!?#6r4 zevg~@t2=llPXbmNt^yMEsa;8>5VyiY2x8-}5rEJrwUc6hiH=L&C)!8h2@sc=WDTr4 zm7s2X93Op$3BH~~M`G$ZZ}>$QK^FS>cb-cNh__vopYO)9utFM=1JwTS)KL9VdEpsI zKq`T9>AWqirjH^7^YWj9e$AGbK+isGjp22;B8ZaHgcG9V(c!BH-canOM~_#b3Cye1 z{c>XA0Vxog*UFOTp*RJf!L=CWX#RwnP4JNQ&Z_E_+<;kqzzulF%|92p-wi}KmwV#- zTY)=rSj=2iICiIspKkB41+7Z-F->Tx`KKEwfxL0($cCT^sr6N8*=)1%%nN#YnJ`hjf{quC zT?uuolWf|?{EaLA$||spa+O_==#HY(!t}pbW5h_dsX(Bi`#PTGfnhGmM$AdOH9~L_ zUqvKqa~{Qu4U*1iI(qXz(Pgr#@c_loBj`yJElm6U& z$OyyUnFTsKLy7-1YyWYsviCyI#U+eX2USXt z@A=O!eQr{kBgBpvkM)%+CQfb3c7;x5LsGQ|ORE(2M?`6o!k41il|&T+O6$L*xXuZY zTi&>plZ5n56f>RH->N-v+!<8OA-fFu*vM@cWupNx;~naEncJvQ2XmG5v4Sr?hYk~E6T8fH+!50 z$uJH2&rm)%{C&hBp+89a)P^KgVY&R|Lb6$2%n%5XOeN2)=%twdHq2Z>jsxbftZn4B6tG#3AC#q~ejK zLsxnX=|$>kKv`gp;dMN^(VA;E2W5P^q{JKY%XV|It2&~KaU_nN(a(v_cri5&4Y2h+ z;of2=x=v3D(2u270qt67$kb$qmPv%<=PFj$zfyf=EIVsRO+*?Sx8l&QEDZXM-$a3& zy`TTuwze9>?F+cmn1@tl8jmAKu0b33^92dTbuv`OPEaKYC_8;BF9YUL_iWAenXHly z!|jA@r%KlA^k3D?+>Tc8r!+-?!bg%w3Hhv;8QdPm(jRIf;#8&m6UewYZ{Z)OP~mKbR8&wQ(iiss)Oo`yPgZ)vOTTXMxDb73Pj`y1tP8sOGbB&6Y5c_qICH-%ch;X4-$!gBkx+* z^mz!xNKBBt`2#6xVU|r(TRxf8Z!=8?P!iLJN0B9sv-8GOWL*So7{ z8#=$muCOn09QLF>->8e@AQOHLy46kri${Pcc*nTgo6GA*k|clClj!n+&hm7@n|QTO z3GY<=p_ENfIIweM&IpR~Dfci5#OeVN6@HRy!L(BoOqqCv)#87McSwAoJ^yrR?J%1d0q{t9ZT^M~p}p%TkI#4Cyc%5&_>#<7E@5Ag+kVv|IJk0P(= z49oTRlA`LquS|P5TP;1Aq5*YMM2{gG2NhuLCF&-}*s5<%3uL`LF1HR!nEkx$zS9Hq z7HG}|u&ktTHT6ZjC}t-zfE2M`uu0g@+2Zc_%!kt~J@^v6PS%WTU~8I)ckYAAM82?T zgtWVK>EqU)LXFH2zRzB@CE-y{m{F^9kohO$c&N_I_3PtK%Fl)v;*h#}1-gJ(_ko-P zsN(>~WWOQ~%ELWu8`NvH;C8EWV}}aAYsz^Xne!;=7|SGhRfyzA-?F{Eslv;bR=!C> zB*eE@x~E9*wzb1+TB9Ja^Yp zROs?O*h6(B(=K!s@o9Sq{>rDowGX-rGFZKTqpo8MA3hymUA8Y?^VnogUrue`qSwby zsFKII^id?gh^On(10hvpigZ4)h_e0Jutt=EvRfG1fm2KWN`G5H9Gg#Ls3}=uz2dd_Am3YKx zhbEG{a8$2ROt1CZoXK`2DMaS?fXopfR-=aCj1+xz@`!M%$TI2z^$D0r*GJIX|`4Z|DV)eC8z8iEYhsiL2I7A$|Pr5wq%A1rtd zN_{P#fHQoO<<29zTBA8v?fP7=sgVqK^sHd#Y@Z~Ky73HKhkg&DsG{#j-_j5^JatRU zV&(=^?`WLa80L0Vs%huI&(~XOl+pa?gZo>XIJ{jO<2!=XP6XM)$k0^EdIV-MKR_Bl zPb^O>Ko3u6Qii)79ql|RF0xk|{rK}jgNcVqscA3y{(A9ct?I(b&% zFZL7q@q|)QQ{#TB(K=`-9p!L(povoEwQF%kcNyw6&7oYMM(~WwpfP!`uxPSZN*PK!uq-AIy*N5 zN7?COAJL_)m!Q>3RkVjzIrbw8ps@W5`D0;E=a}yoe=BQ;8L951H`ZMEjS?Z3Qny0J zy4AmNXCTZ4ArdC~Xfy1F*i8Ze753Nd{oYOHlmPP~-W~mF0^%9+Ir{%V-E*XP2bqH_nm9C;n8Plze|G zPbGZg21$`#y2PQS>#t`Nei;+{c>%)_o$Ljh$UOerT>mO^08J?6JYl8UJRb*$+kJ&Z zel7n`2PsmqSW4hV!2a{dXOuQT?ZQ={}MzG=@4*%=uqg-S+C z04s%_8hnsc`W)eLXG?Q zzK;&#%!WM0H#XwtBD8o)kQ(VjU2)0Z0wvctUgZX{{rRWCOM*^y@s^F#o?{r_` zhWpsUc7Tk{&Y;6WWvZh?{CS7TSdyvDR}O=?0lWP3qiYbV+r5v={>CpdI0N#T@sa>$~u&@OmxT>_iB>)*J`Q!-+c^(z5Blfz~Ob5_)C0A_+_wdj8B z5t^a)&)Or}$Z|K;%@-M_Y**5|tU->=8RI@6jKL%pV&3;PD_p{D*5y+XbmrcEYoDp# ztRFoe!*Az^cD*(+_Cwg$I><13jQT>2!n7$$%{qw9yC(n4Bo@P05XR64?ouc?rn}M@ z(D;ag!2X1uy6ZHFquZNbKV>Y~?C`lxHbrJS_a(E<*h04-p3L4<0Cys6xrZ;>f~F4D zE9iKFx3vE#0q*0=TVVP6Z+?mX)`!~e7*WB%Lvph;kj|6e?zQSa~ZGtyF1xB;l>T^EW^D?BaD{zg(Lsi&a14^D(q# zB4S5T$03HFx?x1jEni&)5S;LrdNF(v@P8Xeo9~$V#Fe-X)O8GeNI~N|e{PsXHMLKx z8^>l(^1e`c;70{=)C*T%E(IO~r5#Gx^+iH$hvg2QDU>GpKLb;LhDeW>z}Ue9s-J3& z7GEHY-(*y3TO<2ZK6ueEN`#tU!0=;W{Nz*F??9;pU54+E<#oQTY;PXn(D~sb6I6_S zi0n`wDC}e3N-({Dwh9wk7WaS*_#rt5ADO>I`%$kfzK0t$Ej+=K_W%FW{|gj=nw@_k zG#`o>3QYt;!UaCVKb|y_KxOLekkF*Q~2MHIbb_(g0DSypiNLXuCoe(wR!D5 zx`C@Ee5((=9r_57Kys~AWx9)D)|RYSy}}K)@;qLD{qbqV6SOE0Fcl$}YL_fB6=jq+E)2~qJqoJ2@nQngeY{|&Yg zA~JidZ`+By&B3vz&rJ9m%l0uH|N8PT|7#Ce5pO(h;79+3{5dtDuO&Xf$7hq6uV-*Y~!{n3__QV9c7f{X59UH)S=#Hw|2VyQU3VRTV7L8@u6VQoAhtVAaQLb&1p{z@3Q=FDerPcx{M?HZQd102gRUi4 z>^TDRBtw`2qDKyitpCGTNj4?$?r0Uus2)phJ7BYQ@SznwGk%17K7<{LG1b;38Sf4&@fpEkJFi1qNh=+5dl76nY z?ap>K6><>qcX&yuX-HgteL~1%$1kc?AmP0|`l`2SfqQ{(d_Z}PEhG@&bD4gROQBsK zpOqHz(4dP>@NFDwuqaNrl5u9xi-S4aa6Y28ezsU%g19<|Z z8vZW*U~V!0Ci>t3uAqODd!T1uwnDn+??Ml-B<}p1#Dh0c%l(_U1FgHiTmQa@zrl6l z-%gJ|0}?dXpL733X+ke8Hzr`*f|2c?w`Edfo&!G-5zQS;(x=sv!T;QSM5&Um%pd=A zCRK#DfhQy0e(SLd9jhl2g$xK$nNlns{{&bA@bN^|Mvy^i68v{WV#+C;5inks24MTsR4A*|MjtS%{sz04YkDC zUo^d%!q55FS75D6wQQO3W)|QhdcLC3cec+xrCv~lSqAqherwo@Md0KE zm)n?QJvwaXDF(h1S-MxP{Ll@;B#SDDlseLdHrCV>j$8V?$G=RcjnziygjAc_ zbXN1R-MDk;Zl;|FqOhDRM5`0cE3cvo*XXaD=MHeYN{T8x@C zH%pYBEM|F;nABU=XSD)v?1%3Za z$`)20FX1frFnxEihyTc#-3PS0Wna*zQGsv#cP#1H>4KF!R)4Og8t|$`+_fH!4~&=P zQO=O~_`Nu%X+uVn+oksbS;C5CsB&ZO%?#Y?c09$z?9w-;OWA!D?5kRKjAy@5Zb^UR zt=XPgQ`HL8eq89*UgOyY0978}s8_5xT=oQXKY$DGFY2n@*LSp~TtsHr)Uj5Dv)D4d zZbz9VU~K$@B7M(kH`zT5*fi<;C=jCYrT!Rq?SEKVJkoM)N{3D|OX#?WTR1dK7BeQlc2J?^)EJ~c9Z1IP)ed|@&zOgoenvqmTxdX(Y;H%37jkmi8nol=u*Sw$yX zjyS|?<#n`bn=rNgJ8je0KQ+!j=k#_%rh<)snx<_!Akz8t6Y&_n?|~j+KYg79WIYbU zw>k%7u>CTj5}ospC<8r~Fm^qhQsPRzGU0CP@W+hdb?4j{lvOn=pI?qkq%uIUu#v2{ zV#HJfoG`O5X02qYUdpKrMYJ03`UHJLmP;yU8KH6nP!GB#?bBWtrtT@aGTR>!G#ueU z*AKtV_$ncnP%kC8ef&jH@0A@^l|Xkz3h_-ET~xklDfZRS*S0DH-NGHog4b%wJJZMO zijsBDh2g#lD`la>^xBwf#;7Bd{x%;PITe)yYpgh2PCi7%ZZY)b>>^MiFKcXjHsXtCuGkC-fdnJ;v2=eD zNTDjg@$3Yhu7*dDT;wAXLOE?-{L#%lMJPzFW*aCXkyYJ4AqQ3xC6gOxS>Q;l&L#gg_F=~HSi|8OX~JiH-kpU8qO3oej{pr)_r7M z+?%uAYV50+11nznD|niHko;9lGQ~!R$#&J__>d>~7f6!4HxFm$6Wr_bkA4cJpUC^nigw>oE2TD;s?R2rTr zY-E5_xui`UBD3uxyJnJm8`oPa5cm>B0fqn{#@7xJvY6RJGhHH!TlmR&9(k|Gce^LP zD&-XWE)&4%YUKwPZS`ZSYg+kHU-8^p!S{M-rjK2bn(uuFCEG6a9_KsCi<^ONOWMey zC<<_X20jYa{(Gv%hq6xTFI#BhHwIw1f{zd#CRKb)bQa@w9smE2EIsGFf)wQ<kNNyQXC?A>_W#)fJ>qEWC*BJbdW{l zEdzflC+S}Fsk*JWyM*^8vai)R!gG3VM%-zwQi^Xd zWgENEdOXU}kP{Xn6Zq0<`C=ejd4#rdn{1e?^-OKH^I5KN1_mjs=PH=gv9UIpi=--jt^Jx2>XZ_MbECjWZk4 zbwKGcO5w(JzsiTQ&&O}nH6osD=HHbNXZ+N%8uU0GJw6mO=d!%eJT~?5*HMy-%g3o) zS~EWJ&0ls3E9Ez?z+WLk#CBbvmc=0OWRhyMY`9QQ$0~RJLo7l#UJP^NQM9_^_pXNm zZps#+YvQ!1FYH{|{EDi$QB$StuEbmR1hx#mJ20c^TPuN|wdcT;sJoSva>h>oOPxbA z(+OKgB=KJe5$|)e8#|o;8R|)(ipo;Bi_|PQVb;ebHAiR%n!``cW-Uxh1dE4AglJN| zz5?#pd~!-7O#4lskA@qjMBJ!rJo*&-Nvu*gM%>>&0WWw5%B-7*T36TkaBmd_yGg*Q zbmL9{Q@`-Foz2Tk3?e7@RJkvU(s;^a#7*|oc6vjIY1zF=f&TKd(Ary11PWjr=J;PH zUY=mDmSu~k zMYsF&PN`^Cxz#LlN}#g`kx^QE)~GD*n{&El z9USg%uel^xXQ6aE*^Nk#VRRGKbjtEkF_-F)O^o^1SDRKmvcRiRaR@dfXo!!Ld(%~o z*ltN>$>xVA)Lv9{G@yFg+k|9qiQhOs6f@ECB9^UmruI?{-_rvMOFi)#+=>@3x&5qR zzvd7uPxV-iihagabSOi)ilcZ?M|H=JNeZL7xn44OcgdtCzhq{Je}tm|J$gPW?sBjU zyftS5W-ppZaGtQAdYSo&KZ~#GdX8m|*Gj~QaPW%1;`d=s^-15=$ud=J*Z$R&2MY6` z)Kk~JHCK)km$}C5@&p8g>)6=+!B*AEzB~E7Wi}c*2;_47;~;SRQ?}Qih6}ff^GejA zS)*ZViklApY*(nMcz) zky{)(@JP6>>Ck%0qh7bF!XG-c9#7vmb*j49ZRWh4qgttbRr9V@zGyBlgDV$3qrBSQ z8>*{N6p)A*blASpA)wl0AFTTUH*70r_GWFP_;qcWev9C;tA zEkp0ddU)e|Ey-vs0atQc^u3IjTjQ|a68PsV;Ye~0{;3Dv^gK_vLfZsxC&Rxcy*Cff;ektwE83KrhLOa#9 ziDtA-Mr4$FUUAk?0g|I`Vzp&m%l8UKa}A3y#)w_y@|<|;?V)&tL|Hj|zN71&be5QR z!PZo2?j$zLW!D_ZK8sVWIlx^+drJ=2PaS_PiE~$bL2Z*sH7{31sn1@4V@Uc*0j@U> z+-rwmtjtkvaO+snwQ#sPo|-2U^Hf~1C*7{AG!(vo(j0-A%B}aDDot@|d;&|uYdNCX zayO^<^YxTxl`>Lm7B*wpY44s(S1}BNUFhTU# z=3PcVGp^`u`gG1QyKpn_V8uF|eTEeJar zHDs+Jty6!+@7Qj;av>r{D{40U(NEK^IYFLbu0HIPOZv@%Qa4;oL?@sk^YXU3^+;AtbI&d^ z(v+4mP;6N{!d{qve?cbwG)hBd_BoddrO?{3SDm(S8(INGnp%(%BR?5s{eM!M$-fwE+AvFrGw|^QwW52$AZCZ2iW0z8p zTlveEcHwi6e-zftTNEgy>gR>H2kVLzm6LKD#wNMPauUx0HTf95$!D{+;xihiP3?c*1-(hr8N(M6K&X zyc0+iat`^apKdTO8Np@HMu!!4SFTlbHh z7s~KnpeVm8Aa+LTs@^-em(4_ubj4|vDDEjlK|g!SGc}x&0PnRA!&NX$zq2SO{?eezy3`%4v0T3x?XrY4?SA{d;TxLKdw zwRNE6+rqh63L5O7yf*{+79z&Xf+wSY{hL5uiI#F#&l|U(*)z^u{cpBJb=CgP;cX{} zV0E4g2^;HHWnB{iJ0&s}8wbXab1b8)Uv>s@Bg%}p>7r6mxfSp7&xu`?@~r*0g_BXF zXKFr6wpm}lOJY6IWL~z!Z$xa{pf(@SnK7KoFm>-tf>R+WdA+5|in;vIXVce+)Tqu9 zCWDUx{PEM_N#=oWp+bz=yoz<@W<7R^esQkm@sG_s<~0YqaopIjaHM%)kz+dG0s1ck z)_p!vkq3*C7NR_IU$_-XE=W&YxjB>HK|!pY7F-$#uL*{ho`C>#vlC057Nrd8>(&}f zA2g~mjv|*Sp%8wuo0qsfs6O*2gWHMsJ4{Q_dn{UBthWKAq^%hCfp(aGU>G9X3g{X&e=5t^JtHtR8hs zoiYY(h1g+&XR4j8FzC(a88 zZsw$u(OhB~-D)@qr3{5wVJs2M&fdChx^7tAoP9=D58#u-As2hYFAU90EOst(sZl;P zVNsG;?t0x2$V4M)%R(JW ziX?G@C5?QnPV!zuTlG7lTeBCtG?X}=2}!6dU3O~T>A+6yS&gZ9e#qErp&J&;D%Lxb^`%U@JmcKmuD&IZ(1*9s>&%!Qh zb>HonU3%(;OblDI*liA6Blr8$bbj(BA4!dI@0#=-*<9pmWoV{<Y0Pr8*fwD5+E!^i-K5-# zdKMzTNUM`gEJ;Mugc;y@Tmua zL?E5>i*nB+g92J z3@J}g*}%qF6ViIMp4S6C1u}M4w&>$KQ_170y1SKeyBi&X{(aXX!%Od@hw^u7gp*|I zyPT1$%sXu@RJx8GJT)l>f_Z7LL$J#WJ3B2#I|KKMD8wSunG8alYvCioci?cHV+%+0 zB{;~8u%t>apTA$mkUdqUe06+Ujfe>Qqj_=R=P1l{bh;tQPCWmPq~TQ}F27XePYqzY zL|Xzb^@(4i{AB3o?PyoTV+|k!`0}Kqv&Xa`(!@&yr^Na&E&LX1gl3{t-pNC*ihUuQ zv$~sw+J^9ipb_{0?JCJq7XqbclZu4YsSiOB) z(4~)UTk^e)z4X~;ktq+Pi1P7tGGh+#4nx$}NOeot9=WAG`BR%(;aMXh^z)1C@{jU} z14MJ=4HD1aZ%LnV*cJDzta*79OhQjz%dO4~s1MJpw_Q&(i{o_sYv`XNTc3JxFj*9K zsc+8-Omu1WZc_=;(1m#AWPiSg@33fs2f%i7{~ zTs<-U%sS-Pbi+fXBd-w-$Ue`>Pd^bLDXYle3hJQ%J7V-|gUHfK?OxYt3#mI_Z}jUL z8{v9AtsgAf7oSyjh_A?Ll_AoSj_t~Ll(H3tv z)n|IiSPA&sy%7vHx=z0eZnGPi8~{HSAJ@;;?$h{!-F6$>Vm6u$-(IfQEz+lT6u~bC z-eLfgpf_=XBmdAjig#c1hW^ecWv{%pm&7*-#LDyF&*BE7{7GhZd~JM!Qv&)s)DVjT0n6r&Ft%tb}hVFm)gCYMD~ zkzstpzZ6mj2rLG+?3jr~)oZYoyX`<4#A*iSe4$WqusM4?@#t>O4&R1p(Rq%MA#1`Q z)~xMryE*9s(Z|_0rSV}pqHf~E4UBlzT(IML* zv6o)vTSo5ab${LzjnPDUx@KeXhq^1^n%yT!;W{UGxRW5liIN>%Pk-^yDE#P=(|g1V zgSKZ<)=Kxu_(Jg->FA?E7)LhfZQoNjnKD(x+ti1gm!4L_Qm$Wiy=e65*!rdn*sMIY zn5g++LDI-i>s2&5x%H-`w$f7%b^ctSO>Dg@jL~EOK4TpdDn%e8$^QEMY@bn#YW-~} zvZ`m;exTZXW#~6*pL%dJ@lJAO6BY`bJUrFpm)_s$Yo3siCcN(?#K%#m)^FY_yiRJx zqa7X812?MaNCJ8ZAo*Vb8^h_!LD>)XrU$uElHHGD24WBVW#C50+Z!`c*dd}3Ne%Te zjdX-KX#pg7cw)qpKRJPF#mIgdu(#h=D&T(LnO`=$z)>tTo6DcTQhSHf^g{y+%Iml` z``i>PDbEfWC4QbpNB`?Q{|NuGrROM4*^cL-+6dQs`Arf)r31$fG8uEdck+Rtfv`Y#3{^d~X!bEBhogy&1aMJ^{;(gf^zo>>_@K3NIm@ zru+tb_MA40mr!WBa@{Y2#C3zguL)@Om8GX7613=y8c1e4exD?4vE@!*--HZdg5B5t z+fHA;YIOo>qjrzCI_}tZd~c1ffAwN-c037CnzL71}=!t zuN*&{-o(`-eLZxun875vw|zp2`7mMS6u%O4kZWX!A4tw)59;zgPVZ*5@)QT>!zPHH zPjYm9YkY<4%ZFjRz@7g2L7X2(Z8B8o%1_{N)?XLxnEk>7zz@ay;WQ5J*>OEGvoACL ziwQzyy=BUV*hpcJ)|rcvAEDPT9zbmM+%Y!lZcg@C&}M9e%5e-C(Zi?7mgh zJVDkDq{W$=2|mmohr>$?(3q?JR5M6`$KE#+yoex}qp-ahdEP9%ZrBA7({SUpp`?fI zO5i}>0R1ize;N<75%$-D-^vwup zAaXU7hG=$z%lBm71*BK6Y_f9ymW(j&gj`3|i}D>Dsjb}mlfh1PTo~ZI*1)8qzxkcM`K# z`}w8KqGVAO-yXu%#Ul}UJ%coFg0acQcGIoa61IjhjfdxYNi{%+mxeMRqi~7lS~m(= zJU;VoIo>D}r7*OrR=_T~l&?g*|LC4bv-I5UZ~^YayBs`sKlqe=ors7s zU}Yq<(lOiu6TGLyEW3XbME_%-Pi=Ix&6*+G25T}X<_dZAQzk52dYbsAY>Y6+?h)&k zFu()DZ(TaWwP1G&$jGe;vKy1ltHiUFMwBKRJSUoRB6H9X?tycGdkDjIH`HO#TX~%s zYm+wdh{*}HBw*NS`&z1TF*|Qoz0u}Nn~U?7C-!{02-(r=j^=9jwr05?H>8w0fW`(? za|yY)iup-PCKVRq)7)kHBLt`O0QK~NPuiMMwMm(Ez;#vjn7LY0uH$4M)!t%M6e;C9-reyf?p0`Kr3mh>2Y>Mtx3&2Y&Ixoz_ zw5KaduK}a#hZaBP@UrP8JGOr$b;N7r*q%bSWP3$2RIPYV`z_u*D?J&2vIcbVWZRN% zNY!WHd02ANnhP4l6Zfl@6l$FO^oI0(O7v8%#?-uZe?~CaC*vA zftjPS%FNb}zRPmPI+~4$sGBr$q-&iltgFI-phvONwBXiJ;Br8gY2Uw08tz>Dmbt)m zO2)#eUNc~DXJ}_4Ze46cc@>;F$|V0F%d-J~C&RvGqRbp~*WBnuLoyLvy3Wmze}dNOspk$KKaxK$n(k z9mVckUa5UouZX9LvsHOCFdWNU(6~OPdJ&$s<8CNDIcDI|ODXSWHKD1<%v_YpktYVJz+1oi z1Ex5dQA1o09Zn?}10wDLHjoD{5{F-tHz&Pq(Zm0T_1Ri%&CT^wTMG$eM44CSf2Tp8EY}%blt+{Iu`R9 zfoVPv4e+s=GP$=-(O%LIaW7HGo+o#R>y_!XBJ>=@{{7=?i_@7qtvwU7th%?s4NWIU zuiLn*-P83a991qb6(|6Ho#R|Mc6zmX;xqbbOvz0z+@KzC@XZ(Hl?YnLZ~>``NDWN8UB5HXWawT zf4z_Cjgt(O@gElWW}4jq|SjGctq4J7J$W8gZJcET!vz~Qe*svpN3!t}YD#x$P6dz9VczrKVg{}hrgME6FYE6;ewU1Fj}A*<{824FU0Ahf{O3w#t5 zv>isl%HCE_&FxE+8gWx`<^Mg|hz!Y{6WQlXh zQL_;mnn`9!HLXUE}s zZm=Kk+Ga2P(17G-EdShG(af7(j5ZV%0*Uos>2k8433fU}SFW#JagO;2?I}{C?sFjU z_|1_j4`T8aJb8(dWQ;a2xR?Z_I=IXSdWok#WRv&ffAVp>UjlsG*3;jGO^%JEu~H0+ zOKfL@6oET9GDUU2*97wtW`cJEfBWIa^~M8a-m^om@?qIeMlQFA@2=A2r7YZKrZ`U8 zRr80Hn_+Fmvr!$J8(PsCS(U#=oVs2#&SRf$+)75z0uAFLZ8tnPj{p04?r$V0zzR}g z;An8GXnx&+)#w>XHO2z}4$JOhi8(BZ{GpdF0kdO-VXs?d2;+}6Bfl=g^p?l_E$3u7 zdc``~Md$i6mq3{+3<&Eu0I8p+z?ebj1v#-9>n2N5OB_9P;^K-p21Eh5a5wt6-OZ4a zDvT9j$j|v*4EPh7rB*Tv29belQ&9eR z{uoe#r*wRe;bSRDWACxKQCGBfL(-W}SUE#MEDa&eep=+I8H|M*+GU+AI6XjKooeqa z&c$$X{#t11N-zDZd{6x0{P%&!2QwFZP0_wFi4uYNf=<)0&*%0c7pd$5t8sgx^ge>2 zktN+jUj6v~Ya zh%=jbY`I==Wf8mjw^7dv%!dH=4D{4mtr+ zD7!bdm2Z3P*-tk@irdj(Y zI3FCVO1^!;5X1oBts2OwY~NZyhi54jDt?hr(@CIWFel9Odsrzb&djx_nJ_JMD3J z{C}jg7>7C8i7pfSz8)w~vH3Md(ll~n{Z=ySbCa@nXcVtrEF?^B=%-^mCGYD&tur55@CTPwxe5lwiwU7ojtG`b9Fn4yu+xs9ev97 z`|zXxnf}Ue`iTFsII!6%q(_F2Gi<9JchF9^yrdi>4R}33g;<{E=%aLZG`&pePH+Y@ z6jC<|eU{HDv9geE*yKVbsK`k921=jI@!Q~~vGLZ|MZloqi~&4TqW6u6y+f_3JtgyZ zxV6Q(68se(KUfwgvC_ngrDx34@r!m%5|~HYm(rAzlj8yN6m$UrJiky57>cQ?pd1NKha!5!;-Jv-FeQK>rf z;BMUGVE1{|nrnFsT0R#<^R^)ThW*IojJT)Gh86J zZJmrtQ65;3YRjPPI-Y>_gnj?ON&-K6SpLlIYHgA8+nT<3z4WB&D9@8*eCVpFX6(}Z z014KYFu_aBsUwa#yCElb2EO9iKJ!>%J1X}ME=5sW_+}_&v!j3p(9H-9mkH%Y;#yce4<^5{C)RH_q9n z%hP_?bjMq)3y$9~Q6s7anUUNF&+zdve;MsYay?VL(htdU&FB%><_x$x;i9*>)8SD3 zWhk0o87Eqz930xsGh`EP!q|W@)hXc(a4K!h2}{yic(<0=DI>MDnofd{?7U-EsFh3k0Jip{`tXcdWv=;-PfciuazY}%QsxS5VcW}D#$HO zVn1QqU=TzS-{3U6Szj(1LoC*?Kzg)~be$hAEB^UGpOz1zfm_Ba=E4LbsJ^+$+?_iw z-W9gm-A{?@7nstBGp^fQU2t}4=(^pv5Rx!0uY}#EII>&UrBH7UfD%GxdEhXQ$E0oH z1oA#syBF|X3f`?w-{#82PS#C}w{G%1>+YAUCcnwmoGwvsU!ZRod(k_vBqxxUbZaF! z&0U!r)sj{BSfjvw9XYrL;|gD}!e9&TY8Eiz(j)S2^l8Az8k}Cb?20v9DROP%;ppNW za9{ z71~h98*neI)XenD`i~EjR>~oL=e8QD>22nwMbkrOJ98S=R2h}~23Lx>^18%zo&8O} zGi4jd4t+CJkG?wI-38|$s85dO^+^NQiIZf|uhIgIY=3(FP9F}y}5b36N zPkk?I=>^!f2C-075TvV&G9w~LkuIVl zy>~DKrRkvJSU~BB5a|dKIzdoSBQ^8@Au^$aUIT=Lcb}7>Fyk}xe9v0n`o14;ez=x7 zVV`|=x%XYJ>pHGhn;Aos=KIqqHQ?pM8B^k4Pv#n})k7Y32tYbO_@PtSwzbuv(4J(B z7eaQllr>;Xk2ppIGYFl|*Cd7oZ0x&~*Br}rzn5_|yd#-ULGEW~TJwkLvO}U0*jP+@Ge654_49!V`l=lfSti+YADU8QI zDC8}?hl3VKyK7oE;dN2FUwZ$$Y`E&eIsoG4QL58R)5Al}Ny*Z{w$ z3*Jbyt{782tGu&fH(6m~rgO61MZ75BKo7_4(a712CsRn3muGUuwPg|&&+-TXv&xcH zOcT2{U)L%0p`|a~bikH<>n`@lnY#lsUeToaa$Ye+gLbh*A>CX>EgJQ*Qz}fAPasf= z`T>ndOoPjrceD(8L@$Rwha(0z<$Ay8mMqLgf%0|Elq9rw`@hVeP_g?w&+Us5B-_rX zyjkQ>QOq9M@=Nro9$U2U5VRu7t({9TT7|I`yXNXvuj3Rb7knu|I)3Tl5@-6>(3@oXs(XwD&ptOmx&C z`*{ANf9T#o6_mX*a%XY*ObaZL;83DUq0nP2SAC5er{@k#5ankdIgw6gC?)5rBIb*#{pZH)8IEr!y4qwn}1 zh!@N19KzS4$F3Ekl=Vz%W#sF&YUfL*kh}`xZb^leX?S;fevT9~uT3#W)_0_e*1<#0 zmAljPA*!NdpmnlWM_D-1acS{i|DC%a5mxcJVh_1sNDs-HfhDz=@nXFYz7?BS&o}ff zIi7UeBe}i`8uF9Iro5APE8fcTn3dS>>^@NB&i1@3+ih>##+z2pQjzv5b+&c@T%@|C z?$lZNBk8`;4chL=oqe{^Xx83=L_7B*`P?#Kn>q6D{0;3d_0)qW@7$Os00eu@Ew|vF z%HpjEdmW2c>hg8H;fM6!(c}VQY<|<=n-vxu17Jd6m0vKNQW84pes9jXt_Svh&WBbK zPuGgb_jHUkQLWvJ%}+(9w)@8_?YBdSjh7o?vr&f&bAVMsMh(S>22gftC(NFG5Ujy8?C z3P=T=j?jS-&bgL0?)VZH%jfKp*u70j^T&Mbc=yV2MBs-6CR@Co1U)gI<+&2+u!H{5 zTv^SHk4V__CUam}9X24^cPzTmY3YXb2NK?MxlAs5DRW*#D^9$|nWcH0(eoa%Yl zudanZ>O6Gx^h9l)m28ks*nFBqVWA@+iQv@jA=>)PYjohi;2w@zT-uI!a=R_xVyyre$!iI<%<=-`S;qYr zP?b5C2~c&xD>}^+p<5#-c^iUg;^YMZV4$BqQ$mh0G=Ue;6rf7Bw6K2%DJk)EO~w+9qU8J(`eR6m-` zu>iLw?`qQ)rXL^UVe7G%>M0slJ-6U(w4Hvxl2N^dLKo#Qf@_1fuy^#d+hg*15@Moj zLxD7{&F}&K)1(&Jns#~2sY16$Z$dnPe7AqDeI&CLJsc2#p!5jtpxYWxP%F1|=SpR1 z;B~b1RxX!JYUA3*FbnT36c)?tyABtgX^3kxxX|A{r^2P33{{$K;YxGp+zV{)@+qHY zRPEB3^GY~c8Vcc0l&52;s>^A~7Dd!jXMw&$J^HP4le#TOD@T78eM7X+<71bLkT3X_ zp9(aX%%8GAI1J}WGM~N5JOo)%JN%IPlCuZjBB>sUPu{CZ(f6c0^o)8V@h{hYDaYo`C;%C{x7x^Q+yB&z0)FpAc_VZhv+4~_SnXYIhJ^Yck94Ff*!QX zcr#Fv%l3YHdxkE%57%f>n!vpT2-4ua5-BxBJ;$yWs|aA)eSCcnXXa%oI4gMdrq)oX zHGQqUpG9Nq4$s(@^*Uzg8J|>ePK(DXhTH0md8X>#*ov69`hl95p}|j;hx5MiI}6#G z$0mhp_6uh;rgLZOZ(1@IHjN-D9}4s<@iW<#q-ZQ-^w=@P@Itb$Y)<4wX_T?Pf{IWF5z}^44 z`*Y6nPg5&y${0slbOm?I8_7D`>k9PC><%d=0de+EFzv^VWAHg$$1EDBW5&AbOU$Xw z1p!gFp7{POlcBIProc#$s?5u`{1L#aR%fK=iZ6R?zQjPbtVQI$KHCPH!{KFF*C-*@ zR_C%E-QJ7MQEx1b_v(MT`quS~HQ3Rwp`l z^Nv?N%BFb}uOwWn!W`osCEe3(97oM1~A#;bw)*MEX(3 z!UjjV+8!)?gSw-W10PS;jSUO3weRs$$*P&@0r-i9w*33M8dzAAcz*r)l)eyBdgdP5 z#f7hJc=9RO3x78p#B4oSct22z$C^5kzAJBw#aC z4A<^9CCno5YGGCunS~o&53V@$YvwZ2iUvr`EZ=wmo8cX*-R0Ezw8`v&snj;;ya9^A zoSzDs{pL68r6vEjYtR|#jw4ve-)qiq~8GUrzuMRHFvpFt_ODL(D-X-iM-f~9G$S8 zrUW?;#}x=5crAtjb25^Ts(!4h(l)VBvGQ^L424yA8>%p~TAVOP84J8LyuX+2%T(Jd z>%h)Cs_niFm}8?@;|m9e^68EzMB&8>G(LocBMWU*q*ceXrTZFB-ZgmXbW)GgXOsGHUfpt)l zakbghelxxnyjqK(8iGu@n3Ef@1m5C>E2rCaGmaj==%Z83ZoL(s6Vb}1F8HHMZe=$! z>%c}wqcbR;Y0IZinHR^Zmd4A3AQah}B=g5UzkBWbP89?q(4eh7u%JQPH7%v|{I^uI z!up@|ulAlV2`fA55n_FfV02xf-2Xz2>K0OpW{>SBbG#Y^;rr^@)Ws2hO%Nf@J(M}J z)i5=qc4bj?`Bl}E_7^T=fl;8U{N>^V2g~Cfh_TQ~=_L#Mj&@LB`Bozck!3jWaznX@R)B!rME5kHeU>qpLlQ0;tf&UuLpeY3E`^F55=1o7g5U4%fXZle zB&8dNiGkuT{;T(SaOYxJFB7cfuzh*lnC^hqNJ{+z%%L_F@CdQyy;z)>!d<6T|W?eh%3n>-TQDxXlmi*oeW}#t^e~eA0&P=G=>EoO`fV zzi1h%FL2nuV=hMl+=CwFCM}$`yMw8nkZFH=vw2D0*iQ*KhlrZ4r-DyrhJO;YVuw0F z>L@|iN@hH_wNZ3Lmd6hGee%Wb@V~B-(AMLS;dq0SBxh$nr#keycpRO-13PFkB3XE9 z#cO8tF$f4oCr&`sOJ*{~eNT+X3$Btb{n|r=V|iIa(a2MM`n8hb)sogWK9|@ZTu7n# z=`|fA6XrsYCWV}Ni)Rhf(b<>>drD|}WB}iBG=Y(vmcHYg5_q(ioWgQ~6P%W_=kHFk zo2fk+XH#SbEdDzHo35T=uin`Q_yLqIHk=Fp;4@_c4Z{GniwKArL*!M)-pU;?j zMPUzo4+qMXJtob{*d%>N0UwE`#N-$>I`%R#F6Jy-THP%p`M_Tny6uXGm5$1_P-h~7@EYt-rH3#w`V)OH4^JV6d@AIzOIwxVzJTE z_qR5j+888OQuQ`Gsmp?EPG#sn&l!z8nRNDpO^fZtBdEafmkHjlFjrdnZz)y71*U#C za-Jk&$br-QV|14Q&&v{jw@vKEQE=T{_Ta15*4iEJI=XC*EctzK8ML(Ey3K;P z=P+8poD-z(_b(r=R%vRjceQ2I|2wuL2(ygJ(Col@GnJShKPCPc@J z)mq9IXN4$f{YutCkn0YgjV^_e_cpIwzsp4l1)Y@} z-ePRm_25h}at6+{R)Hd9!L-pHZFN)vxA*P;uA3JkDq1ii@GNg_+Z9_ z_a&Jiz&)1d+*fdYG*P;qQE=ID)w0lo`)vFkP^hwIWfrQySfV$tEY&ZMsN=}fxH_vA z@}5h{wmq^%N^pSyS`mVch(%S=n(@N)e3m(l$Nz1$sk#Y~FTz8WGXK#td8t+%vH3Ei zqW54%@y6bQUkT5VAo6L2Ok0lvgBNnCk9|L|5!U^#O_W0?guV`|0`HTk#NAHO6>2B> ze_XrgYP*H!%~{V-sQBUXI!!atsOE8zsrJ?YjUG+;ElX?}Z^0PVM=~OGm7Qd&_NY@vL-6C3TfhHzE5K=>O~-m{ z4)Odc=SH>d+50CCGwLrVa!P(O`G~${4osyVyf?(fFL?lLFCa6|7+oho;xb`n7yvr%x7^HBX=Td7;pPX^E{5EGvmt z?5W8vNT2wHuljq~iSN{QzVYd1j0XS&Qm5feLQcp84~3Jl$*}2_7ua&^j_6n>uxsqv zP~uN(ZJPmBw5Cog2n!!{QdOzxBAasFpDlWey_QT{^l7OSJ-OQ9nA(W_oGax1k1%OA zxo%qGZT7XMtyq8V@h^Gp9gNhksT+HhT8J;_+|?)(#6GiGneUj1KVf#D65UC&3?>(B zAiqYKn4dI{X)vtjs49OC9~-M)Y4~L-ny2SGw+dhST4)X|y0l89v%bYebWfTuZiuS(4&%55o6@gXsZ$Q0 zvtY!q(~+4Dn|1`NXq_?31=tr@O?X;r@=7c#@y*kIe>r}au2XdAubUds8rLB-m}FR0 z{qfD7n+3+K6H1Wp%ZYD^y?-#0F!1bT1L&^Gi$rO;Mji(d!7)xw=Q`uQd;M#%LRFDM z0RN))q($^(&5gqJ!!K+~48HueNcxRd_a~Sb%S4I}wQX>QXosmB=Uk80DscEhJKiFu z*26ORx{0pqV_3hzdgnUO6#9BYH3s4yIXGni=-=7C^zWR(e**n5x!t~mVzrxME%ezm zO=K^BJpX#UR_hKyKM>i0dNPj(c=GlyQ*UE8*6;b#w2OoyOEd^OEFG?*B0~KCbFdio z4*lGm-7Cb<%=5_yljpyz8}xCW*KeUn=rM5ql(p~Ks{Wo|$-fR97b4l2JUMaGN9>-K z{hW!9MUnSE1GJcYI{zmo}iP@KrJq{*R-+x%{P3;I_I8XGentmx1;3MDLxz zTC@CVz_(Ok!Yv7+T4eNQe_HLMzoJ&8o~$lQM!|(8_Xre1=j<|nG|;4jyp`@arWciC zOrsqbs44&WBSb;#vA-%V1@27(!CxZqA;$%Mv;jKx4ilYfsd-6=KJ4EzcSJ6wwIOt= zzJOAs`ryupj3$gef}2Wz$t_h=u+f+$st9 zOmmY#3S2^4vypHsA(F<%3#f@RQPoC>j3wSV7ayH}xRdPCqx7vg6ViGqr{6||a!m21 zyRX7wMTCGc{$(^)aHq=A)I|s7>5728MfCe{ztPvlhy?Q_sUH3eM}sK(O>3Wr|8S#! zl}z9sC!v$H&F5!+2gJDfCq6QmqO|n({wAJ>V2N5hsGWhhc$H-16U+WbMoZ6BJ$>ju z4gr`ehoKb6Di?q73j|hQKGrobLNsxD1YT00^h0l710WHE02e+5jLBE$N8!TtER(kVse6GdstaqLIDn29q>I46urfP za{)x3QZO8Uqf9W85?o?oA(3YQsI-K)a9dt5m27p1bIqpAhU6}9?@FzR?G*$A5WAJ* z4i|`KkBju|CBN>hXAuXigxi~|FzE0bpEXL09Nz_sJ1p zf=!g&ya1;EeEq>{Ol{S(mpJj$^>z^N08lL$%xKo|<+%3VlV<0?#REUfkx)G8ZYf)kPF-|WGlizsXbXfYRz3*R zNKbtWKNjUNE0&HCbn@)|TzXY!YQ-H@5whlCFm!exZsq`Fwp_QM9I%#2K7Z^153FQ$ zaslO^h!LpCF=0VCjk+}`Eq~Z`>=Nv_0@8Gk+4#Z^qVCtdsBqUNs`Wz z7zcD(r8aSI6i~7VYz15-?@+^jmH~Fli2GjV$DXhP9qo?IjzofP_?s#>Pk>oEaR5Z+cTY6- z$R02izzyZKfvzO;Z=(XxBt}Yv(%eB-G^icx$nJu3;KqUJr|9)nHLYgew4==9{C9_HKKRTZU9p6~+5S~LTLbl0q`E2mVsTC@Ia%!4~(||(7D4)L7z@=cS{gOp(cf7zq53KTu8-sy9(8W zEpzW%zA(t%mCwa_f8ztagzVEep5l5tbJkm>6 zqW$c1UWhCHH;bZc8yVdmRr^l2_b**t@Ow3{Djk;GaAK9^EAZ~rG=$CG7U85*IQu#l zOwIh`8RF9;z0Ibr#(`UcIKM#4JYNFrAk;TmBsexd;Pqz!kj4(}ixMM+khe+=Vkl$Rd(SB!)% zM$>0es_rZQk&d9EE=zqM! zHpJjrJ5^FTY%n6KPr$yZ%o`#f1`-dFB?bs_+%6M#kg{Gg4d)o_<|h~U>8%yCdhhO? zHxR>x6>iH^zi1ATG7okCtg31^W6sfNoQ*LjnQZ%}dq6gin`pO7ClY+)XlW(WXFsma zzdN#I&d#LH;sR>$757Gbq`t2OJh)}0K$~!Fs zgrbPJS}%A>tyVWbxjAl1t~*_d=6h6#XZ^kbbU^Zk=Lh2ymq(O>vSKftvgk~8$;>FX z_1&LP-(ruB+Gl3kVPvW;Zr*2v=N9jgz1D{B9^7YrX1qC1nqKTpqq&)SB&eC+xN&K< z7m{~_$nUj;paXn^sNk41OaWnGN$tM{2%95@Raw1{A+{+n9Eptc?4Ww2F5idq!$1WF z=>G8EJ069~pC-%2S)vjEL!ItU)y>w#$>TM_%i(NQ;ylK#wH?ONhPb%A9rM)!C}9Bu z3rEpbUsVKzw{f)306*lYNY9l?u+ROJF7~iaL_tnzp5%I?dQ!abm%SXMhWgz1zzVb z^=?~%&IQc2+7jXpMVT2|J#IwZBE3!Po)ieY0IF%~8}5;=@u!9++J{lRD*<5XUg+`= zt;idb9oJki>~(kfFk9LJv1ju(t?sMfcdS4LBfUffBz6`+e99rmb7kWAs$2zPFkXi; zf}_7#Jm<47_FwCFS2rI&AV;U&uT!{qD<O;509cRDS}`6{wQ#_F9e&O7&HAU;y@-6N6s+2?g zy4vI{xOeeNv9I6v0+1Zgqunk5D19q-z;Qz~Hpy>&w?1pvfPb?`WVk{x$>x|I%jW0j|v9!SN9 zwa)0OsrRo;P4DHuTzAXL{>{jvB=7G?ANA^3PunVW-IQwKzm*aCy&+h~90B+2`0)!% z02M?G;81i?h5XGee@NsL*XaSdzmR#Hz+Esdr6w}Y^SvFk4&4n-=E``WMpf&cSls&T z2KueXx}l5K0ZT#Z$9?KD>V@3+5;CR3AiwWUtF89 zL&fV9ke--vTW}R*%UN*pm7brH`j6V>4L(t1(NX40P?O(-Z~#8%M7^=sZ|kP%!ds2g zk{B#X-JIQH;!j=&;`dWuGl)09P@XvQWiJw>jscIE=Nvd&+WTEmY%RbZedC;!9`-mp zMbg1K&iF0-;-mAIY5$f@1xf)dLPx+!AvKn zu-d;mDEQBF&F-4T;LbCZ_qKA-8OkF)C+C2%ys8}EV0?DHau$lD3q;M_VD%#r*|qb$ zN-hR{atXb&g#}{x-sIH*@G}$T3MK2;?_?*qor}B^VLiCj4!%|vvFnvm;}>~HTMk|4 z2Cfyxd$_O{22y{xMV3;F&KKIDM%X=5{u1qTQ$@tNZhw{9yQjkt;-2@E^~A7*Y8@^W z&L3Fwu02PXtX_B$`oXDCPMo{ix3oW*);XECNSi6%^R*eTBJ#Ih(O5>&xAiDCPoW&OD&@1cq1~bzEMQkU4Z>-J7){(qiBS~AJz`g?Ap-|>I>Q=LJLq)yKMBAtYx*gUcKlx(>cBON@bKJ;(98@v#hNTsW6SKUz& z&(;}4i&LAkcH$H?JMj8RDff!%@~pEc4Tu9OT^^9nUU8M08(JWgm<>tOY78S#t}|_I zQTtkz=O*G83kXU}fVn!kKzT9JoNS`uPK&0H2sB!*ascO^ zIP9_C>TYBTO)yWGO*~}@1{sp$KyhgoXO{QvK%~t zn>I^)IWJx@nQnf`-QP|&qKkYoD^>%m>#tQD5Ta|vBDORA=}eSe#r8VWiLkCub3@*w zGO@1uX$715!)BXaJx0_iMOJ0r2!EaBe(=%Sx z%3}*#M~LeC?ycMSYnI9t@#SRXazh0DJz=S`Nty;2_w-Jhs2O4bg^aTNJ>Qr8*@hPa zbr$I31}^$v>JtwaYRtw-UJ5njBzMaekkE(v?9fpR=GaCj$5Td8r1_FT%?8)|ZA}gn z`7bB(5)cg_zs=oVV;eLJf<%<*<|3Iz6X|+qpOVagQoA^*%kCrjWc24jsaKrEMsF1mv(|f6LC_ z87F2AfI7m6q%;YXcM=I=t^r86b;V4N27x}0c{Ra0*qdq60m?zp4rYXKYm7r+TqC!D zojF*3_+UAbv7OXbH?nS&Sl#?lJdV4$Mpr7MJE|hTuSEZZ;;hKcDoOQyH}JWpe05YzIMi>u%7!0w6aydkNMZ!fb>}E>|9Be+L*=1 zhnlHgzbcK)D$4~yG&3r?@ZoEi(ZqLcu<8iG}W8Wnva|JPLR)m^+ zTC`OgX0B*c&6D%#(fGs!PlYBxUD2h!81A08<+t)7OKF>^A{}$iIL^>-BN_@fG^1@d zt(;tQ-u-NbF|L9hQ|WC8HxMq+EY4`G*=YKp!e#soJ~=Vl*Z8v(fa{WM@ShxQ)LAXu z`Rn$!Bq0t(S=38oU2=HFtlQ`9@m(%6+=#K@wy{XyM6Y?euOiVMn_-i?T8@B#536Il z;fHL3geCZ2U|b(8_PVM7fM)I~=!YOtBEWYH=xDCi?jU3n;{xi=0zY-OyuuV3&oW8l><{zk^om#h~*E)7mQoCK{ypTPG_OEQGw zUbgE&!0)t%F6<6rEVnCrN=B`3UQoS6*#Ju{flEfh%txY?{dlk8gE-k5`UCI@Y-Oex z8tVW{WoGIwA5bWQKY<2J!_rvCWeK`bF*-QV3Ib)wXWIi^$s198N*TZ~d0em6r}9VpH*t?dc-qC_AzTkkT$#F03L!NNPts^3;GBxUokvrML2alN+k-`($_)WdHfGX4ixq8>E4c1k`t^zu6i zuNA>{G6jO<;8(y!`Gt7~ekR^#e_p3lou&Rwy3V@x?;c}aR>JXjSlS%4_#-ngmI>n$ zxGp!rUC6suzO7L5`!fxV?!Ir;u>AhN8=W;8cwXITpiB-w%DUxtEXF(3Gz{d8zUra; z!ytBQU|`W$0!KjK5MaluH}t%QDSxzg+A6Nc>cT{5zduGN_%A&d9H_MdVsth3Z#MnI zI1aoyWD{->w*Zp(yOese5q)np#YdLWcqb*^#emt`fawLwQgQsa`NU%Hp3Eidu3-)v8`RZ06oqMXh=?#N#6d4;Zw+hXvAg2&ua2o zK&f%sEjC&k?qlR( z7v0I1iD|r5cDrnzVs&y;hOfKx&rMj(^!g%*2?2(Mnpzf8ImPBC-U^rv<$hhqPTEUp z$W<4+75j-`(Iz2&BoEHE7N(G++igQa?5q+?PW9d||?%c5oZB<99lY(ONwb$82sb3+wJ{4s1` zwzFD{_1Tg^pZ1(;TS@BGOqOl%r{Mw9PCk1=)Uai4rRO+Bx#bbAN1tSRx;Es9RblpF z`Af6?Oggz&JouYSvdYd|N0HnI2l(?5>X*ZCCiCx&Sbu;&@xsjB9XVJX(Y$ICP@e5X z0ua#uZBpBj)ahi7mbCsH@X76Y-K6$RyCZ^XltHCD#X6uPIxCJgOc`M-79!|VYHCL> zc@1ul#Osh#&-@I1-eoEC5oBQd&Kq%T`9K|WL%W+$RF-inV`Q&1f5P#g6AITwYerF- zsHb(FcJz`gXHk3Vb*{FW9`EQl2)cOA_0mSIS$^Dr&7oHIvhmv5!d4TUqiew^YFfJHU)0_ zr2MM6g=Wc~fBUcUb=m5uqi^-wOJbrvwjjnod8K(Cm*(g~Z&^a9XjwcvfVHwyshgj! z@NeGer#nLyLC$$VVNq#-@P+@V`;q&8_rQ%RdI;<|KbJ+-#LPWSo+uhDy^!jfPJ^h! zj}uDgPqCT<@+-PD7M+o$AN7MI%26*VVZK&UeSFFDL>8cF8MUw7!tkRs-v4`J%NWWM zS8(3`jivqTsC-Rie>1a;riN-Z^8B5RW&EmV^;G;frxg6FL_t2;zdNEBO|4x9tJ11# z&nNg-fljMu?7umj;9vb;87!vw+BXyRr)OizihVP^f9eWDEbaTl%zEXlnI!z868Mjx ze;DKc|M0>Ze~!^9u(&`_gq1M-+DG7Pg82x%GjeRzQG_q?My4;Vu=+WyJko34MTEZk N^~{Bzvrk{S{XbBdgf##F literal 0 HcmV?d00001 diff --git a/examples/img/sequential_pipeline.png b/examples/img/sequential_pipeline.png new file mode 100644 index 0000000000000000000000000000000000000000..12168e3b35503d6c38f3ef73b8c9fe42277a77d0 GIT binary patch literal 8359 zcmd^EXIN8Pvj!9p5U^0CDWFJ=2q=i5NS7)l5?W{>^s4lt0wTSGv>=KiO)x-^t_T5w z5TpmB8xROmq=lBdIePAMeth@)b${HlpJXz#*50#c&Ae-7){|%hJq0_vIhvHo?_aY#iL@4VjQ&7B5WH_*)2Ifu~=oqU3 z<2={Zl)Z9!r8EHk_cgP!;};tFw?y%zYHDS3@3;68vj8xsS$TeoAHNJsnmi=!9|B_A z_!>YfdY67rIeYp73?&uSZ&Cc(b%6P`doa1YliD-~@HHR107@p40jlJY4N%^T{5^OA z!2G`C9pZHijL&cj$=W;xMv}8%{vPj{-e5j+=Fd%k{ghNU)eo%ztVhw9xSk8uN$T;% zZ9r5|K3<@C&JaK&;!**+utqYsaR8X#*C1h&LsTAFRDs#vumOfl9vJEw9ouAas+&wc zDVbcQ)1X5WiveCh=cx(xp`c(td;F!uUiqa6EW1NXRoU3zY9%jTMXJLK zGhWM=#9{cF5h5({PWQ}wQ?+%GYp^jDE0NHS-*gI02)SJ<2pWuuEBzEN_N)8T zvwdf0oO%&1T(UNM@W|0QYJg~)-Q&=2ouP|3??zGzU0TQt5fu{gGw}$3bXlxeEAkP2 zOuoNlwsd&nIJp%3c){4FJHGH9y0Zb&z?S{CuYAm{fJ<~*i`D-HC(GynOYWr9m#7@QqZh`x%VP4g*=etXeAvx1+eTjtGh29k&%z>Sm{teJbGZ>ke;5 z-MD)ZQWSDUHb++~tHR+rL*se6<+de|+=}V_?afRB_Wd4iSVCb-V(_qB>STmbJ85`61})sNDxp%*h~jmREl zGKgf;i%qm(9-nm|%kO=&!#*F>;-?7i1NSe~$NQvNmOP#(+I!qu%o4AWmP?un8Xy*4 zlSfpMJia3?A+|CLzd4H0?YuqAO01wkXrLNv5vuE3Pt zL!82osMy_{_5;*LZR}<9E!PY%-dlVxJ};$bz9l6;61g&S0<$=|*_KhbtsLKf=V(M5 zoRi4Dr>sa-Rn}0Gt_ImD3k|#iGBm3Hs>I+FV6{oC#5X;xmV}Yq(`*m3C*V-BZN1;u zksFA;4itXV6jIfkr)t(k4t$TGol`#Y2Y0^K|9TmbsFCsz7jC{%lA|$BYc*W8Ju`m= zi)3BccD*FA+!5NkB~w$Z*SGwRI|6QdC{{hr5yoJ(F&m@kFLR#=x1 zu^)H;b(NQ)FHXPR$zCSl$W5hp(rMKU$Cf{B=J#qQ1+9`WDMHjdRORAn<=Nj0cvogj zzG`QJqiPIh6)!xrw?g2nqPFq1j7B|p`EKYT+jrM1qOx5_5tgH_M^oPs?-8$i9l(p9 zweVHvYKW>wwd$hPNg(MhSA}Mqu8`2i6d7Hc?)hzS0}|c;qbpa9cGd1Zsv15IkE8_pvXAbWO?u&)ub*xO z0P+u*e6ImLYCRvll6+tTUwhh}Rzo?>bK}U{5hvw(KVUJ)rkBC(@PlUg)AB*9XL z?&Hp53n)<%tosh|cJ0yHc-|Z%TPKhG#(TzxScuC-cwNf!%`XqS%?4rUetIWE3*S~y z8=K0dQ6Ho4Bc8(}VN}MYgA>N+UbtF|>+pb3X|Cta1O1E?TK#(QoQ-o(3u|q0cn7LP zXzATujO=+(aI*wFs0qn}xA6@NlQ54$6*!4CZ!uLJ6@;q!1il|x&O;0YfuWJeXMP)F zVmZ^IT8y$8xaSf!!a;8qy>8qB_w(=MDt1E@6lyBmx^eI$+2+6`sg6DF3x|QWq4L9F z%4WS|Vdw+r>C3_{WWCBGBdnqC*S^Y?kkYLdU>1JD!0F ztG{@iU1eCsrivD>_Gh5bPacLi7(D*`Jgn;4ESqo%u}Scym`>yCvY+g=7uT$>Ncc=7 zj7FnTR2>_f1D7%0o1dCv#e`qNgdsHf()uxbkb~6u(b~j8lyN_6H;D$Vji-jhKe^#p z1-gO@JJ4Zckkl9!FD$b*&q%wl1?J|`+tV@3F}`Pq9kUw0q$pxBZ2x5pOV8ak5|$nGaN~U}!DO{%C=GSl1E%obB7hWmlrmunqqbKYlfUE6n(JJKeK4Ln zJ9~$=ay?Te2x&71irPMCMNf%6`20HyAVzn)3&1Ufk5l|zu0;b==c9S7&ygbCgwDxd zq-%?ngvshcxhR`$#+hCyw^X{(_Psbx92Dce6pVY3Zt{<44pR($X-Cl~9Q#17f6sg-Ni)PQ(vkzoyM*CjfC-W=+r}wxMPgUqm>`}>Vp`d?LZdnC3_=Pv z8k~kWZR6grsina9F&xIQw+HTK51(?_i3R^3R})eydL+D7?ls{Q7YRx^I46mDv3G}1 z>+vuIyJY7)Hd8~B^p%KDfDU8d@n+Ii&AZNNs~_St`s8NFj_?+gNXC=fV1B!-g5982A^$<4!(WC z+YRfGiu@Qqdr4p@O3p-XY7VRZt&{xhH4vZ7PQH#>(Tz`rXq|?Jw}mqZUwxPj<5{2J)F;({F-SKAlebBmT0{+i_q%AlNF`zZW`(G@(NZ#zLPWIbYB%fl8L@|^L$?j?(Egh8ef z)@mtSS-;}@%5FceV`*Y}dgdh?b<(bp0!RL$lX>| zDLk&=r{N78c|SkX5Ry7@j;Iut=d)}zR~qRU=q}WijlAk~@v=?=c0QGu8F`3K8@ls) zM1GrU-0g#LK6xcOji6*RfUZnynhy6jsk?5(xAN)5-tB>C?vm4$7v2%1-fE9pKpcwg zLI&T@M;~GU?g4Kpnj4nVSnSV7uQEt3%G#o~i+&e)~9@)mE1nE;sj zRL^hLJ>Pz3s-TZ|PH$tsH&T9nqTHO&i~W@mn>M7YMQy%R`bcGKswRtrn_s5D$Ow|b zk(>QS-``vbi`6FuFS35fz`c;Tq@j`hz@J52k8kdMe)g-BYyMbo8MsW1bggub!5gk5 zHa|VQ#@PZKO=i0$f_4POpCjG}fqh{fqib<|-YueOsM1~RCb_ew9rnvU_e#3z~N6z223o=U%UGHd)yJ%T1o2>UUrn&&l5X_sA{s84%9sJ z^=g{sBgUUkQuR`wYq!4B#*+j-CZqPm;N$(YY@qwlM?!AP#oEFRNmL2nU0TuF&@U%+ z%)xLt^i5f>o2%{Go^UDMxB^_93tn8MsdqYNK_pYSl-AAmj0^UNNgDu~SdlAka`7R( zXak6NM>}zAt5(f?0p*8&j#b7B7>&s{6=sPS62FOf@snXk^xXi+{9)+KyxdX$@qy(M zGpyd`g!1{CqE+fZ(q)y6AtN@r@mp|@je&KcEWASZbp#eW{;SF&R4YY6am>lC;UERC zS|)6M5yKHnNE1hzfxI$aTV+jG{v4tMt#Ys=e(W$>ApvJ#>?~V&GcQmSFPA~TdLnL0 ztEN1|=t?y1>+c87_#+t;w^mg7LKIi>5K=rvAr^OU{FQ-JkLa|A``1m59u=?r(DWH` z_Xf`+0-@W_r>3i4Nf9f7lDNys3h_L0sT{2i@!6ZgA|VnI+y@o{aLK>Mx+sG;-bWE)L*5#70sg*9%?i0 z!B#)Rj$Wa-u*vYJkheg8Jca$Uio2Vw&}{qgCu{mgQ2zicC7zE~Z50qxTBQuk)ufnj z4|>EKbuzu_#19cg(`T%*UDHobMl><}@X|EE|2<`xyuxxSh`nzv=>=+s0?b4w5=2|b zKQ4S`TjE8>B}$uvWwhzz)J<8zi*W+dhSPQ;s5~kgZl?ECLW^FSN*pE_`rN|f`?w2| zsrZbJmCbnquhL3Xp=y159WpHls5~`yDxPKcd5mk35=WFu*+;EF?sI3<)Bo0?;aK>w z!`NP)x>c6jmRhfWF6q&=lDI!QWbX)2E4(>22s3j=)H4NGP{cj>A03#p7lhX&S@C+0 zuQ5gaZKjCb9~}vOnvWBUJALURyky({m}$)UM~7F0I(@K?8rtY$+~-Sw%zTCYTgU&i zrOuUp2uw}ZYx;!fRqx5LQ zUo}prG~`hTaemVy#c3ZS(1d9(%S%>n8u?SM3uX7tof^Pl-Uh*=j_P8Zi1~nueH`zO zjN}&-=lzD00H*XvQx(Tk0@bW`!!*jyYA&V;n?vZPm_v)Dbo1%CQa1cyuwZ?D$39Bw zUV30oqF%?GhSwZS^ivS)C)YLZ$N0uja?FE_Z zDqIRXC%qnu0*L>~!h>u%c)|s`igMxKc!}M+0Ove5;p-&Xl+6s-4Q`QXT_NUmtlA^! z;ip?8mpL%pd4s$%Zg)$(dFWossqub%)w7a?e7 zTyHRPVxX0!i)v`U1+y-QyjJ3>D}s`5?FG?`M|ee`ZLREUu-@g=U)iC(nz3L)L~chg z8=g<(V5=kjV7+_jz+I;0$oKM8FnB@!N;bQw)E&U zfZ_LDpc1H^eie0ns(TmrV}Ci|llJ%7*>AI3f)HoQ%NamOBnyvOV|z?UQyJ8?fqM25 zm?=77)h7b|QVeAXexU1X(VY&u+f~fT=I>!x)PP;|UJuBiLtd}vv9Q%W!{GBP)4?d9 z+AV-kWfjEmYQ^2IwPVNF{;Te#^|^dEW0^pLkz4^SwVjlF&SYa#~I9 zzsGR1mN|w!9h9NTZ`xlwO!C%=sV;ZZm~O(I9g7KIO`EmJhj%3$oX*5($SF?LHMhR! zvQmdxb!pWb8gmF=gFLO^i`BI}6&X-m^nnm0JTw|pa>J)k)Yhd`M3jksGFHrk-J&~N zf=;Q=q@`}ZU^ zQ0VawPPC3cQCwn#QUVFOWKi9^DJi2Ggu2CP6&x@y3YN_%9hx1h{$>ibhoh*ONC}mv za;3xMX`0AY9$}HXr(R))2k+Qz?&0esx{tnD*y92k;Gx)q_6^_mpMH|5tka&grVFn= zik!g!uzaYt?Ybzhldcl0wmtAB%ast3gg}*Mm2NTomA0Oe6j8@_&KSTglf^4`qPgc* zsIo5L7X%5<(p&~-V_;TA0{3aI!9qbQjSg|#kPhh^5LPHYu?Hv~f!~A`zmrKuY^Kz$ zZ&5PZ1#%yYV$~|RH`tt2{JNNYWj+fGz{VR>cZAHdT1;1`UShG5C;6Z9%Y!O`q7P(2 z%j9&E$vhs5@h^IBynUBVJjx)kWs~;_U}%F)z^#@ZGt*R5tcWP6ealw_CH^S&w0qIP zo{$CQU-mi)f>@y9sfR^Y6@HB)wU325>`x&#H&Lt=C2^YIPUdD=IdIyi5Lfx?iy%&} zvEjU20b2GOOD5qxqe8bztTBpOrc}piznJB>`TdMOgRn)O69fKr76zV^Px;M*=L%up z5W+BvNiErhH1xUt@<)GKt!mcg7~=FEjTD?Rv#lM*6`ZmhHhzyKe-_82{Xi3#KrnldqSIH+|QvGfM|7CY&gSaJuIz@-0?eTF2x*5E? aM*b9AKV;#g&v5+Tx|W)r>bpDEPyPc`C?XI5 literal 0 HcmV?d00001 diff --git a/sktime/forecasting/model_evaluation/_functions.py b/sktime/forecasting/model_evaluation/_functions.py index cec248d5f98..8ae6e774190 100644 --- a/sktime/forecasting/model_evaluation/_functions.py +++ b/sktime/forecasting/model_evaluation/_functions.py @@ -252,7 +252,7 @@ def _evaluate_window(x, meta): # make prediction if y_pred_key not in y_preds_cache.keys(): start_pred = time.perf_counter() - y_pred = method(fh, X_test, **pred_args) + y_pred = method(fh=fh, X=X_test, **pred_args) pred_time = time.perf_counter() - start_pred temp_result[time_key] = [pred_time] y_preds_cache[y_pred_key] = [y_pred]