diff --git a/imbens/ensemble/base.py b/imbens/ensemble/base.py
index 5e67ae0..f63cb67 100644
--- a/imbens/ensemble/base.py
+++ b/imbens/ensemble/base.py
@@ -42,8 +42,13 @@
     check_random_state,
     column_or_1d,
     has_fit_parameter,
-    validate_data,
 )
+try:
+    from sklearn.utils.validation import validate_data
+    HAS_VALIDATE_DATA = True
+except ImportError:
+    HAS_VALIDATE_DATA = False
+
 
 TRAINING_LOG_HEAD_TITLES = {
     "iter": "#Estimators",
@@ -516,6 +521,27 @@ def _validate_y(self, y):
         self.n_classes_ = len(self.classes_)
         return y
 
+    
+    def _validate_data(self, X="no_validation", y="no_validation", reset=True, validate_separately=False, **check_params):  
+        """Cross-compatible _validate_data for scikit-learn < 1.6.0 and >= 1.6.0"""
+        if HAS_VALIDATE_DATA:
+            return validate_data(
+                self,
+                X=X,
+                y=y,
+                reset=reset,
+                validate_separately=validate_separately,
+                **check_params
+            )
+        else:
+            return super()._validate_data(
+                X=X,
+                y=y,
+                reset=reset,
+                validate_separately=validate_separately,
+                **check_params
+            )
+
     def _validate_estimator(self, default):
         """Check the estimator, sampler and the n_estimator attribute.
 
@@ -572,7 +598,7 @@ def fit(self, X, y, *, sample_weight=None, **kwargs):
         self.random_state = check_random_state(self.random_state)
 
         # Convert data (X is required to be 2d and indexable)
-        X, y = validate_data(self, X, y, **self.check_x_y_args)
+        X, y = self._validate_data(X=X, y=y, **self.check_x_y_args)
         if sample_weight is not None:
             sample_weight = _check_sample_weight(sample_weight, X, dtype=np.float64)
             sample_weight /= sample_weight.sum()
@@ -704,9 +730,8 @@ def decision_function(self, X):
         check_is_fitted(self)
 
         # Check data
-        X = validate_data(
-            self,
-            X,
+        X = self._validate_data(
+            X=X,
             accept_sparse=["csr", "csc"],
             dtype=None,
             ensure_all_finite=False,