MAINT Replace stats.mode calls with fixes._mode

build_tools/azure/install.sh

@@ -68,7 +68,7 @@ elif [[ "$DISTRIB" == "conda-pip-latest" ]]; then
     python -m pip install -U pip
 
     python -m pip install pandas matplotlib
-    python -m pip install --pre scikit-learn
+    python -m pip install scikit-learn
 
 elif [[ "$DISTRIB" == "conda-pip-latest-tensorflow" ]]; then
     make_conda "python=$PYTHON_VERSION"

imblearn/over_sampling/_smote/base.py

@@ -12,7 +12,6 @@
 
 import numpy as np
 from scipy import sparse
-from scipy import stats
 
 from sklearn.preprocessing import OneHotEncoder, OrdinalEncoder
 from sklearn.utils import check_random_state
@@ -29,6 +28,7 @@
 from ...utils._docstring import _n_jobs_docstring
 from ...utils._docstring import _random_state_docstring
 from ...utils._validation import _deprecate_positional_args
+from ...utils.fixes import _mode
 
 
 class BaseSMOTE(BaseOverSampler):
@@ -786,7 +786,7 @@ def _make_samples(self, X_class, klass, y_dtype, nn_indices, n_samples):
         # where for each feature individually, each category generated is the
         # most common category
         X_new = np.squeeze(
-            stats.mode(X_class[nn_indices[samples_indices]], axis=1).mode, axis=1
+            _mode(X_class[nn_indices[samples_indices]], axis=1).mode, axis=1
         )
         y_new = np.full(n_samples, fill_value=klass, dtype=y_dtype)
         return X_new, y_new

imblearn/under_sampling/_prototype_selection/_edited_nearest_neighbours.py

@@ -9,7 +9,6 @@
 from collections import Counter
 
 import numpy as np
-from scipy.stats import mode
 
 from sklearn.utils import _safe_indexing
 
@@ -18,6 +17,8 @@
 from ...utils import Substitution
 from ...utils._docstring import _n_jobs_docstring
 from ...utils._validation import _deprecate_positional_args
+from ...utils.fixes import _mode
+
 
 SEL_KIND = ("all", "mode")
 
@@ -155,7 +156,7 @@ def _fit_resample(self, X, y):
                 nnhood_idx = self.nn_.kneighbors(X_class, return_distance=False)[:, 1:]
                 nnhood_label = y[nnhood_idx]
                 if self.kind_sel == "mode":
-                    nnhood_label, _ = mode(nnhood_label, axis=1)
+                    nnhood_label, _ = _mode(nnhood_label, axis=1)
                     nnhood_bool = np.ravel(nnhood_label) == y_class
                 elif self.kind_sel == "all":
                     nnhood_label = nnhood_label == target_class

imblearn/under_sampling/_prototype_selection/_neighbourhood_cleaning_rule.py

@@ -7,7 +7,6 @@
 from collections import Counter
 
 import numpy as np
-from scipy.stats import mode
 
 from sklearn.utils import _safe_indexing
 
@@ -17,6 +16,8 @@
 from ...utils import Substitution
 from ...utils._docstring import _n_jobs_docstring
 from ...utils._validation import _deprecate_positional_args
+from ...utils.fixes import _mode
+
 
 SEL_KIND = ("all", "mode")
 
@@ -182,7 +183,7 @@ def _fit_resample(self, X, y):
         nnhood_idx = self.nn_.kneighbors(X_class, return_distance=False)[:, 1:]
         nnhood_label = y[nnhood_idx]
         if self.kind_sel == "mode":
-            nnhood_label_majority, _ = mode(nnhood_label, axis=1)
+            nnhood_label_majority, _ = _mode(nnhood_label, axis=1)
             nnhood_bool = np.ravel(nnhood_label_majority) == y_class
         elif self.kind_sel == "all":
             nnhood_label_majority = nnhood_label == class_minority

imblearn/utils/fixes.py

@@ -0,0 +1,20 @@
+"""Compatibility fixes for older version of python, numpy, scipy, and
+scikit-learn.
+
+If you add content to this file, please give the version of the package at
+which the fix is no longer needed.
+"""
+
+import scipy
+import scipy.stats
+
+from sklearn.utils.fixes import parse_version
+
+sp_version = parse_version(scipy.__version__)
+
+
+# TODO: Remove when SciPy 1.9 is the minimum supported version
+def _mode(a, axis=0):
+    if sp_version >= parse_version("1.9.0"):
+        return scipy.stats.mode(a, axis=axis, keepdims=True)
+    return scipy.stats.mode(a, axis=axis)