Fix plot_lm and _z_scale for scipy 1.10 (arviz-devs#2186)

* stop using np.tile on plotter list * take nan_policy into account for scipy>=1.10 * fix version comparisons to take prereleases into account * update changelog * workarounds to be able to merge
TimOliverMaier · Dec 22, 2022 · fadbc20 · fadbc20
1 parent 3c82bdd
commit fadbc20
Show file tree

Hide file tree

Showing 4 changed files with 25 additions and 6 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -11,6 +11,8 @@
 - plot_bpv smooth discrete data only when computing u_values ([2179](https://github.com/arviz-devs/arviz/pull/2179))
 - Fix bug when beanmachine objects lack some fields ([2154](https://github.com/arviz-devs/arviz/pull/2154))
 - Fix gap for `plot_trace` with option `kind="rank_bars"` ([2180](https://github.com/arviz-devs/arviz/pull/2180))
+- Fix `plot_lm` unsupported usage of `np.tile` ([2186](https://github.com/arviz-devs/arviz/pull/2186))
+- Update `_z_scale` to work with SciPy 1.10 ([2186](https://github.com/arviz-devs/arviz/pull/2186))
 
 ### Deprecation
 

diff --git a/arviz/plots/lmplot.py b/arviz/plots/lmplot.py
@@ -1,6 +1,7 @@
 """Plot regression figure."""
 import warnings
 from numbers import Integral
+from itertools import repeat
 
 import xarray as xr
 import numpy as np
@@ -11,6 +12,10 @@
 from .plot_utils import default_grid, filter_plotters_list, get_plotting_function
 
 
+def _repeat_flatten_list(lst, n):
+    return [item for sublist in repeat(lst, n) for item in sublist]
+
+
 def plot_lm(
     y,
     idata=None,
@@ -268,8 +273,8 @@ def plot_lm(
     len_y = len(y)
     len_x = len(x)
     length_plotters = len_x * len_y
-    y = np.tile(y, (len_x, 1))
-    x = np.tile(x, (len_y, 1))
+    y = _repeat_flatten_list(y, len_x)
+    x = _repeat_flatten_list(x, len_y)
 
     # Filter out the required values to generate plotters
     if y_hat is not None:
@@ -289,7 +294,7 @@ def plot_lm(
             )
         ]
 
-        y_hat = np.tile(y_hat, (len_x, 1))
+        y_hat = _repeat_flatten_list(y_hat, len_x)
 
     # Filter out the required values to generate plotters
     if y_model is not None:
@@ -307,7 +312,7 @@ def plot_lm(
                 ),
             )
         ]
-        y_model = np.tile(y_model, (len_x, 1))
+        y_model = _repeat_flatten_list(y_model, len_x)
 
     rows, cols = default_grid(length_plotters)
 

diff --git a/arviz/stats/diagnostics.py b/arviz/stats/diagnostics.py
@@ -4,7 +4,9 @@
 from collections.abc import Sequence
 
 import numpy as np
+import packaging
 import pandas as pd
+import scipy
 from scipy import stats
 
 from ..data import convert_to_dataset
@@ -538,7 +540,13 @@ def _z_scale(ary):
     np.ndarray
     """
     ary = np.asarray(ary)
-    rank = stats.rankdata(ary, method="average")
+    if packaging.version.parse(scipy.__version__) < packaging.version.parse("1.10.0.dev0"):
+        rank = stats.rankdata(ary, method="average")
+    else:
+        # the .ravel part is only needed to overcom a bug in scipy 1.10.0.rc1
+        rank = stats.rankdata(  # pylint: disable=unexpected-keyword-arg
+            ary.ravel(), method="average", nan_policy="omit"
+        )
     rank = _backtransform_ranks(rank)
     z = stats.norm.ppf(rank)
     z = z.reshape(ary.shape)

diff --git a/arviz/tests/base_tests/test_diagnostics.py b/arviz/tests/base_tests/test_diagnostics.py
@@ -3,8 +3,10 @@
 import os
 
 import numpy as np
+import packaging
 import pandas as pd
 import pytest
+import scipy
 from numpy.testing import assert_almost_equal
 
 from ...data import from_cmdstan, load_arviz_data
@@ -488,9 +490,11 @@ def test_nan_behaviour(self, func):
         data[0, 0] = np.nan  #  pylint: disable=unsupported-assignment-operation
         if func == "_mcse_quantile":
             assert np.isnan(_mcse_quantile(data, 0.5)).all(None)
-        else:
+        elif packaging.version.parse(scipy.__version__) < packaging.version.parse("1.10.0.dev0"):
             assert not np.isnan(_z_scale(data)).all(None)
             assert not np.isnan(_z_scale(data)).any(None)
+        else:
+            assert np.isnan(_z_scale(data)).sum() == 1
 
     @pytest.mark.parametrize("chains", (None, 1, 2, 3))
     @pytest.mark.parametrize("draws", (2, 3, 100, 101))