Fix Improve error when categorical_features is an empty list (scikit-learn#31146)

Signed-off-by: Pedro Lopes <pedro.m.a.r.lopes@tecnico.ulisboa.pt>
Co-authored-by: Jérémie du Boisberranger <jeremie@probabl.ai>

Author: pedroL0pes

doc/whats_new/upcoming_changes/sklearn.inspection/31146.fix.rst

@@ -0,0 +1,4 @@
+- :func:`inspection.partial_dependence` now raises an informative error when passing
+  an empty list as the `categorical_features` parameter. `None` should be used instead
+  to indicate that no categorical features are present.
+  By :user:`Pedro Lopes <pedroL0pes>`.

sklearn/inspection/_partial_dependence.py

@@ -673,6 +673,12 @@ def partial_dependence(
         is_categorical = [False] * len(features_indices)
     else:
         categorical_features = np.asarray(categorical_features)
+        if categorical_features.size == 0:
+            raise ValueError(
+                "Passing an empty list (`[]`) to `categorical_features` is not "
+                "supported. Use `None` instead to indicate that there are no "
+                "categorical features."
+            )
         if categorical_features.dtype.kind == "b":
             # categorical features provided as a list of boolean
             if categorical_features.size != n_features:

sklearn/inspection/tests/test_partial_dependence.py

@@ -1196,3 +1196,22 @@ def test_reject_pandas_with_integer_dtype():
         warnings.simplefilter("error")
         partial_dependence(clf, X, features=["a"])
         partial_dependence(clf, X, features=["c"], categorical_features=["c"])
+
+
+def test_partial_dependence_empty_categorical_features():
+    """Check that we raise the proper exception when `categorical_features`
+    is an empty list"""
+    clf = make_pipeline(StandardScaler(), LogisticRegression())
+    clf.fit(iris.data, iris.target)
+
+    with pytest.raises(
+        ValueError,
+        match=re.escape(
+            "Passing an empty list (`[]`) to `categorical_features` is not "
+            "supported. Use `None` instead to indicate that there are no "
+            "categorical features."
+        ),
+    ):
+        partial_dependence(
+            estimator=clf, X=iris.data, features=[0], categorical_features=[]
+        )