MAINT replace deprecated scipy distance_matrix with cdist

scipy 1.18 deprecates scipy.spatial.distance_matrix in favor of
scipy.spatial.distance.cdist (removal scheduled for scipy 1.20), which
currently raises a DeprecationWarning (and fails the SMOTEN tests that
treat warnings as errors). Switch ValueDifferenceMetric to cdist with the
Minkowski metric; the computed distances are identical.

Author: chaoz23

doc/whats_new/v0.15.rst

@@ -17,5 +17,12 @@ Enhancements
 Compatibility
 .............
 
+- Replace the deprecated ``scipy.spatial.distance_matrix`` with
+  ``scipy.spatial.distance.cdist`` (Minkowski metric) in
+  :class:`~imblearn.metrics.pairwise.ValueDifferenceMetric`. This silences a
+  ``DeprecationWarning`` under SciPy 1.18 (the function is scheduled for
+  removal in SciPy 1.20); the computed distances are unchanged.
+  :pr:`1188` by :user:`chaoz23`.
+
 Deprecations
 ............

imblearn/metrics/pairwise.py

@@ -6,7 +6,7 @@
 import numbers
 
 import numpy as np
-from scipy.spatial import distance_matrix
+from scipy.spatial.distance import cdist
 from sklearn.base import BaseEstimator
 from sklearn.utils import check_consistent_length
 from sklearn.utils._param_validation import StrOptions
@@ -227,7 +227,8 @@ def pairwise(self, X, Y=None):
             else:
                 proba_feature_Y = proba_feature_X
             distance += (
-                distance_matrix(proba_feature_X, proba_feature_Y, p=self.k) ** self.r
+                cdist(proba_feature_X, proba_feature_Y, metric="minkowski", p=self.k)
+                ** self.r
             )
         return distance