do not use mixin but implement functionality

Author: glemaitre

imblearn/model_selection/_split.py

@@ -1,12 +1,14 @@
+import warnings
+
 import numpy as np
 from sklearn.base import clone
 from sklearn.model_selection import LeaveOneGroupOut, cross_val_predict
-from sklearn.model_selection._split import BaseCrossValidator, _UnsupportedGroupCVMixin
+from sklearn.model_selection._split import BaseCrossValidator
 from sklearn.utils.multiclass import type_of_target
 from sklearn.utils.validation import _num_samples
 
 
-class InstanceHardnessCV(_UnsupportedGroupCVMixin, BaseCrossValidator):
+class InstanceHardnessCV(BaseCrossValidator):
     """Instance-hardness cross-validation splitter.
 
     Cross-validation splitter that distributes samples with large instance hardness
@@ -72,6 +74,12 @@ def split(self, X, y, groups=None):
         test : ndarray
             The testing set indices for that split.
         """
+        if groups is not None:
+            warnings.warn(
+                f"The groups parameter is ignored by {self.__class__.__name__}",
+                UserWarning,
+            )
+
         classes = np.unique(y)
         y_type = type_of_target(y)
         if y_type != "binary":

imblearn/model_selection/tests/test_split.py

@@ -21,6 +21,16 @@ def data():
     )
 
 
+def test_groups_parameter_warning(data):
+    """Test that a warning is raised when groups parameter is provided."""
+    X, y = data
+    ih_cv = InstanceHardnessCV(estimator=LogisticRegression())
+
+    warning_msg = "The groups parameter is ignored by InstanceHardnessCV"
+    with pytest.warns(UserWarning, match=warning_msg):
+        list(ih_cv.split(X, y, groups=np.ones_like(y)))
+
+
 def test_error_on_multiclass():
     """Test that an error is raised when the target is not binary."""
     X, y = make_classification(n_classes=3, n_clusters_per_class=1)