scikit-learn
diff --git a/‎doc/whats_new/v1.3.rst
Copy file name to clipboardExpand all lines: doc/whats_new/v1.3.rst
+3Lines changed: 3 additions & 0 deletions b/‎doc/whats_new/v1.3.rst
Copy file name to clipboardExpand all lines: doc/whats_new/v1.3.rst
+3Lines changed: 3 additions & 0 deletions
diff --git a/‎sklearn/preprocessing/tests/test_label.py
Copy file name to clipboardExpand all lines: sklearn/preprocessing/tests/test_label.py
+6-2Lines changed: 6 additions & 2 deletions b/‎sklearn/preprocessing/tests/test_label.py
Copy file name to clipboardExpand all lines: sklearn/preprocessing/tests/test_label.py
+6-2Lines changed: 6 additions & 2 deletions
diff --git a/‎sklearn/utils/tests/test_validation.py
Copy file name to clipboardExpand all lines: sklearn/utils/tests/test_validation.py
+19Lines changed: 19 additions & 0 deletions b/‎sklearn/utils/tests/test_validation.py
Copy file name to clipboardExpand all lines: sklearn/utils/tests/test_validation.py
+19Lines changed: 19 additions & 0 deletions
diff --git a/‎sklearn/utils/validation.py
Copy file name to clipboardExpand all lines: sklearn/utils/validation.py
+12-1Lines changed: 12 additions & 1 deletion b/‎sklearn/utils/validation.py
Copy file name to clipboardExpand all lines: sklearn/utils/validation.py
+12-1Lines changed: 12 additions & 1 deletion
@@ -407,6 +407,9 @@ Changelog
   :pr:`25733` by :user:`Brigitta Sipőcz <bsipocz>` and
   :user:`Jérémie du Boisberranger <jeremiedbb>`.
 
+- |FIX| Fixes :func:`utils.validation.check_array` to properly convert pandas
+  extension arrays. :pr:`25813` by `Thomas Fan`_.
+
 :mod:`sklearn.semi_supervised`
 ..............................
 
 
@@ -118,15 +118,19 @@ def test_label_binarizer_set_label_encoding():
 
 
 @pytest.mark.parametrize("dtype", ["Int64", "Float64", "boolean"])
-def test_label_binarizer_pandas_nullable(dtype):
+@pytest.mark.parametrize("unique_first", [True, False])
+def test_label_binarizer_pandas_nullable(dtype, unique_first):
     """Checks that LabelBinarizer works with pandas nullable dtypes.
 
     Non-regression test for gh-25637.
     """
     pd = pytest.importorskip("pandas")
-    from sklearn.preprocessing import LabelBinarizer
 
     y_true = pd.Series([1, 0, 0, 1, 0, 1, 1, 0, 1], dtype=dtype)
+    if unique_first:
+        # Calling unique creates a pandas array which has a different interface
+        # compared to a pandas Series. Specifically, pandas arrays do not have "iloc".
+        y_true = y_true.unique()
     lb = LabelBinarizer().fit(y_true)
     y_out = lb.transform([1, 0])
 
 
@@ -1762,6 +1762,25 @@ def test_boolean_series_remains_boolean():
     assert_array_equal(res, expected)
 
 
+@pytest.mark.parametrize("input_values", [[0, 1, 0, 1, 0, np.nan], [0, 1, 0, 1, 0, 1]])
+def test_pandas_array_returns_ndarray(input_values):
+    """Check pandas array with extensions dtypes returns a numeric ndarray.
+
+    Non-regression test for gh-25637.
+    """
+    pd = importorskip("pandas")
+    input_series = pd.array(input_values, dtype="Int32")
+    result = check_array(
+        input_series,
+        dtype=None,
+        ensure_2d=False,
+        allow_nd=False,
+        force_all_finite=False,
+    )
+    assert np.issubdtype(result.dtype.kind, np.floating)
+    assert_allclose(result, input_values)
+
+
 @pytest.mark.parametrize("array_namespace", ["numpy.array_api", "cupy.array_api"])
 def test_check_array_array_api_has_non_finite(array_namespace):
     """Checks that Array API arrays checks non-finite correctly."""
 
@@ -626,6 +626,15 @@ def _pandas_dtype_needs_early_conversion(pd_dtype):
     return False
 
 
+def _is_extension_array_dtype(array):
+    try:
+        from pandas.api.types import is_extension_array_dtype
+
+        return is_extension_array_dtype(array)
+    except ImportError:
+        return False
+
+
 def check_array(
     array,
     accept_sparse=False,
@@ -777,7 +786,9 @@ def check_array(
         if all(isinstance(dtype_iter, np.dtype) for dtype_iter in dtypes_orig):
             dtype_orig = np.result_type(*dtypes_orig)
 
-    elif hasattr(array, "iloc") and hasattr(array, "dtype"):
+    elif (_is_extension_array_dtype(array) or hasattr(array, "iloc")) and hasattr(
+        array, "dtype"
+    ):
         # array is a pandas series
         pandas_requires_conversion = _pandas_dtype_needs_early_conversion(array.dtype)
         if isinstance(array.dtype, np.dtype):