introduced our own NotFittedError, update check_is_fitted, minor updates

sktime · mloning · Jun 4, 2020 · Jan 23, 2020 · Jan 27, 2020 · Jan 27, 2020
commit 65914252c822ee8843b6b40c6b42f8a5fca793ac
diff --git a/sktime/classifiers/compose/ensemble.py b/sktime/classifiers/compose/ensemble.py
@@ -5,29 +5,29 @@
 __all__ = ["TimeSeriesForestClassifier"]
 __author__ = "Markus Löning"
 
-
-from warnings import warn
 from warnings import catch_warnings
 from warnings import simplefilter
+from warnings import warn
+
 import numpy as np
+from joblib import Parallel, delayed
+from sklearn.ensemble.base import _partition_estimators
 from sklearn.ensemble.forest import ForestClassifier
 from sklearn.ensemble.forest import MAX_INT
 from sklearn.ensemble.forest import _generate_sample_indices
 from sklearn.ensemble.forest import _generate_unsampled_indices
 from sklearn.ensemble.forest import _get_n_samples_bootstrap
-from sklearn.ensemble.base import _partition_estimators
-from sklearn.utils._joblib import Parallel, delayed
+from sklearn.exceptions import DataConversionWarning
+from sklearn.tree import DecisionTreeClassifier
 from sklearn.tree._tree import DOUBLE
-from sklearn.utils import check_random_state
 from sklearn.utils import check_array
+from sklearn.utils import check_random_state
 from sklearn.utils import compute_sample_weight
-from sklearn.utils.validation import check_is_fitted
-from sklearn.exceptions import DataConversionWarning
-from sklearn.tree import DecisionTreeClassifier
 
 from sktime.pipeline import Pipeline
 from sktime.transformers.summarise import RandomIntervalFeatureExtractor
 from sktime.utils.time_series import time_series_slope
+from sktime.utils.validation import check_is_fitted
 from sktime.utils.validation.supervised import validate_X_y, check_X_is_univariate, validate_X
 
 

diff --git a/sktime/classifiers/distance_based/time_series_neighbors.py b/sktime/classifiers/distance_based/time_series_neighbors.py
@@ -17,7 +17,7 @@
 from sklearn.model_selection import GridSearchCV, LeaveOneOut
 from sklearn.utils import gen_even_slices
 from sklearn.utils.multiclass import check_classification_targets
-from sklearn.utils.validation import check_is_fitted
+from sktime.utils.validation import check_is_fitted
 from sklearn.utils import Parallel, delayed, effective_n_jobs
 from sklearn.utils._joblib import __version__ as joblib_version
 from sklearn.exceptions import DataConversionWarning

diff --git a/sktime/contrib/rotation_forest/rotation_forest_reworked.py b/sktime/contrib/rotation_forest/rotation_forest_reworked.py
@@ -10,7 +10,8 @@
 from sklearn.exceptions import DataConversionWarning
 from sklearn.preprocessing import StandardScaler
 from sklearn.tree import DecisionTreeClassifier
-from sklearn.utils.validation import check_X_y, check_array, check_random_state, check_is_fitted
+from sklearn.utils.validation import check_X_y, check_array, check_random_state
+from sktime.utils.validation import check_is_fitted
 
 
 class RotationForestClassifier(ForestClassifier):

diff --git a/sktime/exceptions.py b/sktime/exceptions.py
@@ -0,0 +1,16 @@
+#!/usr/bin/env python3 -u
+# coding: utf-8
+
+__author__ = "Markus Löning"
+__all__ = ["NotFittedError"]
+
+
+class NotFittedError(ValueError, AttributeError):
+    """Exception class to raise if estimator is used before fitting.
+    This class inherits from both ValueError and AttributeError to help with
+    exception handling and backward compatibility.
+
+    References
+    ----------
+    ..[1]   Based on scikit-learn's NotFittedError
+    """
diff --git a/sktime/regressors/tsf.py b/sktime/regressors/tsf.py
@@ -15,7 +15,7 @@
 from sklearn.utils import check_array
 from sklearn.utils import check_random_state
 from sklearn.utils._joblib import Parallel, delayed
-from sklearn.utils.validation import check_is_fitted
+from sktime.utils.validation import check_is_fitted
 
 from sktime.classifiers.compose.ensemble import _parallel_build_trees
 from sktime.transformers.summarise import RandomIntervalFeatureExtractor
@@ -400,7 +400,7 @@ def decision_path(self, X):
 
     @property
     def feature_importances_(self):
-        raise NotImplementedError
+        raise NotImplementedError()
 
     def _set_oob_score(self, X, y):
         raise NotImplementedError()
diff --git a/sktime/transformers/compose.py b/sktime/transformers/compose.py
@@ -7,7 +7,7 @@
 import pandas as pd
 from scipy import sparse
 from sklearn.compose import ColumnTransformer as skColumnTransformer
-from sklearn.utils.validation import check_is_fitted
+from sktime.utils.validation import check_is_fitted
 
 from sktime.transformers.base import BaseTransformer
 from sktime.utils.data_container import concat_nested_arrays

diff --git a/sktime/transformers/forecasting.py b/sktime/transformers/forecasting.py
@@ -1,6 +1,6 @@
 import numpy as np
 import pandas as pd
-from sklearn.utils.validation import check_is_fitted
+from sktime.utils.validation import check_is_fitted
 from statsmodels.tsa.seasonal import seasonal_decompose
 
 from sktime.transformers.base import BaseTransformer

diff --git a/sktime/transformers/pca.py b/sktime/transformers/pca.py
@@ -3,7 +3,7 @@
 
 import pandas as pd
 
-from sklearn.utils.validation import check_is_fitted
+from sktime.utils.validation import check_is_fitted
 from sklearn.decomposition import PCA
 
 from sktime.transformers.base import BaseTransformer

diff --git a/sktime/transformers/segment.py b/sktime/transformers/segment.py
@@ -1,7 +1,7 @@
 import numpy as np
 import pandas as pd
 from sklearn.utils import check_random_state
-from sklearn.utils.validation import check_is_fitted
+from sktime.utils.validation import check_is_fitted
 
 from sktime.transformers.base import BaseTransformer
 from sktime.utils.data_container import check_equal_index, tabularize, concat_nested_arrays

diff --git a/sktime/transformers/summarise.py b/sktime/transformers/summarise.py
@@ -1,6 +1,6 @@
 import numpy as np
 import pandas as pd
-from sklearn.utils.validation import check_is_fitted
+from sktime.utils.validation import check_is_fitted
 
 from sktime.transformers.base import BaseTransformer
 from sktime.transformers.segment import RandomIntervalSegmenter

diff --git a/sktime/transformers/tests/test_PCATransformer.py b/sktime/transformers/tests/test_PCATransformer.py
@@ -1,5 +1,5 @@
 from sklearn.decomposition import PCA
-from sklearn.exceptions import NotFittedError
+from sktime.exceptions import NotFittedError
 from sktime.transformers.pca import PCATransformer
 from sktime.utils.testing import generate_df_from_array
 from sktime.utils.data_container import tabularize, detabularize, get_time_index

diff --git a/sktime/utils/validation/__init__.py b/sktime/utils/validation/__init__.py
@@ -0,0 +1,75 @@
+__all__ = ["check_is_fitted"]
+__author__ = "Markus Löning"
+
+from inspect import isclass
+from sktime.exceptions import NotFittedError
+
+
+def check_is_fitted(estimator, attributes=None, msg=None, all_or_any=all):
+    """Perform is_fitted validation for estimator.
+
+    Checks if the estimator is fitted by verifying the presence of
+    fitted attributes (ending with a trailing underscore) and otherwise
+    raises a NotFittedError with the given message.
+
+    This utility is meant to be used internally by estimators themselves,
+    typically in their own predict / transform methods.
+
+    Parameters
+    ----------
+    estimator : estimator instance.
+        estimator instance for which the check is performed.
+
+    attributes : str, list or tuple of str, default=None
+        Attribute name(s) given as string or a list/tuple of strings
+        Eg.: ``["coef_", "estimator_", ...], "coef_"``
+
+        If `None`, `estimator` is considered fitted if there exist an
+        attribute that ends with a underscore and does not start with double
+        underscore.
+
+    msg : string
+        The default error message is, "This %(name)s instance is not fitted
+        yet. Call 'fit' with appropriate arguments before using this
+        estimator."
+
+        For custom messages if "%(name)s" is present in the message string,
+        it is substituted for the estimator name.
+
+        Eg. : "Estimator, %(name)s, must be fitted before sparsifying".
+
+    all_or_any : callable, {all, any}, default all
+        Specify whether all or any of the given attributes must exist.
+
+    Returns
+    -------
+    None
+
+    Raises
+    ------
+    NotFittedError
+        If the attributes are not found.
+
+    References
+    ----------
+    ..[1]   Based on scikit-learn's `check_is_fitted` function
+    """
+    if isclass(estimator):
+        raise TypeError("{} is a class, not an instance.".format(estimator))
+    if msg is None:
+        msg = ("This %(name)s instance is not fitted yet. Call 'fit' with "
+               "appropriate arguments before using this estimator.")
+
+    if not hasattr(estimator, 'fit'):
+        raise TypeError("%s is not an estimator instance." % (estimator))
+
+    if attributes is not None:
+        if not isinstance(attributes, (list, tuple)):
+            attributes = [attributes]
+        attrs = all_or_any([hasattr(estimator, attr) for attr in attributes])
+    else:
+        attrs = [v for v in vars(estimator)
+                 if v.endswith("_") and not v.startswith("__")]
+
+    if not attrs:
+        raise NotFittedError(msg % {'name': type(estimator).__name__})