scikit-learn · thomasjpfan · May 5, 2023 · May 4, 2023 · May 5, 2023 · May 5, 2023
diff --git a/doc/whats_new/v1.3.rst b/doc/whats_new/v1.3.rst
@@ -459,6 +459,10 @@ Changelog
   `pandas.DataFrame`
   :pr:`25220` by :user:`Ian Thompson <it176131>`.
 
+- |Fix| :meth:`pipeline.Pipeline.fit_transform` now raises an `AttributeError`
+  if the last step of the pipeline does not support `fit_transform`.
+  :pr:`26325` by `Adrin Jalali`_.
+
 :mod:`sklearn.preprocessing`
 ............................
 

diff --git a/sklearn/pipeline.py b/sklearn/pipeline.py
@@ -421,6 +421,14 @@ def fit(self, X, y=None, **fit_params):
 
         return self
 
+    def _can_fit_transform(self):
+        return (
+            self._final_estimator == "passthrough"
+            or hasattr(self._final_estimator, "transform")
+            or hasattr(self._final_estimator, "fit_transform")
+        )
+
+    @available_if(_can_fit_transform)
     def fit_transform(self, X, y=None, **fit_params):
         """Fit the model and transform with the final estimator.
 
@@ -744,12 +752,34 @@ def classes_(self):
         return self.steps[-1][1].classes_
 
     def _more_tags(self):
+        tags = {
+            "_xfail_checks": {
+                "check_dont_overwrite_parameters": (
+                    "Pipeline changes the `steps` parameter, which it shouldn't."
+                    "Therefore this test is x-fail until we fix this."
+                ),
+                "check_estimators_overwrite_params": (
+                    "Pipeline changes the `steps` parameter, which it shouldn't."
+                    "Therefore this test is x-fail until we fix this."
+                ),
+            }
+        }
+
         try:
-            return {"pairwise": _safe_tags(self.steps[0][1], "pairwise")}
+            tags["pairwise"] = _safe_tags(self.steps[0][1], "pairwise")
         except (ValueError, AttributeError, TypeError):
             # This happens when the `steps` is not a list of (name, estimator)
             # tuples and `fit` is not called yet to validate the steps.
-            return {}
+            pass
+
+        try:
+            tags["multioutput"] = _safe_tags(self.steps[-1][1], "multioutput")
+        except (ValueError, AttributeError, TypeError):
+            # This happens when the `steps` is not a list of (name, estimator)
+            # tuples and `fit` is not called yet to validate the steps.
+            pass
+
+        return tags
 
     def get_feature_names_out(self, input_features=None):
         """Get output feature names for transformation.

diff --git a/sklearn/tests/test_common.py b/sklearn/tests/test_common.py
@@ -133,7 +133,17 @@ def _tested_estimators(type_filter=None):
         yield estimator
 
 
-@parametrize_with_checks(list(_tested_estimators()))
+def _generate_pipeline():
+    for final_estimator in [Ridge(), LogisticRegression()]:
+        yield Pipeline(
+            steps=[
+                ("scaler", StandardScaler()),
+                ("final_estimator", final_estimator),
+            ]
+        )
+
+
+@parametrize_with_checks(list(chain(_tested_estimators(), _generate_pipeline())))
 def test_estimators(estimator, check, request):
     # Common tests for estimator instances
     with ignore_warnings(category=(FutureWarning, ConvergenceWarning, UserWarning)):
@@ -283,16 +293,6 @@ def _generate_column_transformer_instances():
     )
 
 
-def _generate_pipeline():
-    for final_estimator in [Ridge(), LogisticRegression()]:
-        yield Pipeline(
-            steps=[
-                ("scaler", StandardScaler()),
-                ("final_estimator", final_estimator),
-            ]
-        )
-
-
 def _generate_search_cv_instances():
     for SearchCV, (Estimator, param_grid) in product(
         [

diff --git a/sklearn/tests/test_pipeline.py b/sklearn/tests/test_pipeline.py
@@ -670,7 +670,8 @@ def test_set_pipeline_steps():
     with pytest.raises(TypeError, match=msg):
         pipeline.fit([[1]], [1])
 
-    with pytest.raises(TypeError, match=msg):
+    msg = "This 'Pipeline' has no attribute 'fit_transform'"
+    with pytest.raises(AttributeError, match=msg):
         pipeline.fit_transform([[1]], [1])