FIX Allows pipeline to pass through feature names (#21351)

thomasjpfan · web-flow · commit a3f09ea82dd3 · 2021-10-19T10:10:22.000+02:00
diff --git a/doc/whats_new/v1.0.rst b/doc/whats_new/v1.0.rst
@@ -60,6 +60,13 @@ Fixed models
   and `radius_neighbors`, due to handling of explicit zeros in `bsr` and `dok`
   :term:`sparse graph` formats. :pr:`21199` by `Thomas Fan`_.
 
+:mod:`sklearn.pipeline`
+.......................
+
+- |Fix| :meth:`pipeline.Pipeline.get_feature_names_out` correctly passes feature
+  names out from one step of a pipeline to the next. :pr:`21351` by
+  `Thomas Fan`_.
+
 .. _changes_1_0:
 
 Version 1.0.0
diff --git a/sklearn/pipeline.py b/sklearn/pipeline.py
@@ -746,15 +746,16 @@ def get_feature_names_out(self, input_features=None):
         feature_names_out : ndarray of str objects
             Transformed feature names.
         """
+        feature_names_out = input_features
         for _, name, transform in self._iter():
             if not hasattr(transform, "get_feature_names_out"):
                 raise AttributeError(
                     "Estimator {} does not provide get_feature_names_out. "
                     "Did you mean to call pipeline[:-1].get_feature_names_out"
                     "()?".format(name)
                 )
-            feature_names = transform.get_feature_names_out(input_features)
-        return feature_names
+            feature_names_out = transform.get_feature_names_out(feature_names_out)
+        return feature_names_out
 
     @property
     def n_features_in_(self):
diff --git a/sklearn/tests/test_pipeline.py b/sklearn/tests/test_pipeline.py
@@ -1517,3 +1517,24 @@ def fit(self, X, y):
         check_is_fitted(pipeline)
     pipeline.fit(iris.data, iris.target)
     check_is_fitted(pipeline)
+
+
+def test_pipeline_get_feature_names_out_passes_names_through():
+    """Check that pipeline passes names through.
+
+    Non-regresion test for #21349.
+    """
+    X, y = iris.data, iris.target
+
+    class AddPrefixStandardScalar(StandardScaler):
+        def get_feature_names_out(self, input_features=None):
+            names = super().get_feature_names_out(input_features=input_features)
+            return np.asarray([f"my_prefix_{name}" for name in names], dtype=object)
+
+    pipe = make_pipeline(AddPrefixStandardScalar(), StandardScaler())
+    pipe.fit(X, y)
+
+    input_names = iris.feature_names
+    feature_names_out = pipe.get_feature_names_out(input_names)
+
+    assert_array_equal(feature_names_out, [f"my_prefix_{name}" for name in input_names])
