FIX fix regression in gridsearchcv when parameter grids have estimators as values (#29179)

MarcoGorelli · jeremiedbb · commit 6428b980e3ca · 2024-07-02T19:14:07.000+02:00
diff --git a/doc/whats_new/v1.5.rst b/doc/whats_new/v1.5.rst
@@ -38,6 +38,10 @@ Changelog
   grids that have heterogeneous parameter values.
   :pr:`29078` by :user:`Loïc Estève <lesteve>`.
 
+- |Fix| Fix a regression in :class:`model_selection.GridSearchCV` for parameter
+  grids that have estimators as parameter values.
+  :pr:`29179` by :user:`Marco Gorelli<MarcoGorelli>`.
+
 
 .. _changes_1_5:
 
diff --git a/sklearn/model_selection/_search.py b/sklearn/model_selection/_search.py
@@ -1089,9 +1089,24 @@ def _store(key_name, array, weights=None, splits=False, rank=False):
         for key, param_result in param_results.items():
             param_list = list(param_result.values())
             try:
-                arr_dtype = np.result_type(*param_list)
+                with warnings.catch_warnings():
+                    warnings.filterwarnings(
+                        "ignore",
+                        message="in the future the `.dtype` attribute",
+                        category=DeprecationWarning,
+                    )
+                    # Warning raised by NumPy 1.20+
+                    arr_dtype = np.result_type(*param_list)
             except (TypeError, ValueError):
-                arr_dtype = object
+                arr_dtype = np.dtype(object)
+            else:
+                if any(np.min_scalar_type(x) == object for x in param_list):
+                    # `np.result_type` might get thrown off by `.dtype` properties
+                    # (which some estimators have).
+                    # If finding the result dtype this way would give object,
+                    # then we use object.
+                    # https://github.com/scikit-learn/scikit-learn/issues/29157
+                    arr_dtype = np.dtype(object)
             if len(param_list) == n_candidates and arr_dtype != object:
                 # Exclude `object` else the numpy constructor might infer a list of
                 # tuples to be a 2d array.
diff --git a/sklearn/model_selection/tests/test_search.py b/sklearn/model_selection/tests/test_search.py
@@ -17,6 +17,7 @@
 from sklearn import config_context
 from sklearn.base import BaseEstimator, ClassifierMixin, is_classifier
 from sklearn.cluster import KMeans
+from sklearn.compose import ColumnTransformer
 from sklearn.datasets import (
     make_blobs,
     make_classification,
@@ -64,7 +65,7 @@
 from sklearn.naive_bayes import ComplementNB
 from sklearn.neighbors import KernelDensity, KNeighborsClassifier, LocalOutlierFactor
 from sklearn.pipeline import Pipeline
-from sklearn.preprocessing import StandardScaler
+from sklearn.preprocessing import OneHotEncoder, OrdinalEncoder, StandardScaler
 from sklearn.svm import SVC, LinearSVC
 from sklearn.tests.metadata_routing_common import (
     ConsumingScorer,
@@ -1403,9 +1404,7 @@ def test_search_cv_results_none_param():
             est_parameters,
             cv=cv,
         ).fit(X, y)
-        assert_array_equal(
-            grid_search.cv_results_["param_random_state"], [0, float("nan")]
-        )
+        assert_array_equal(grid_search.cv_results_["param_random_state"], [0, None])
 
 
 @ignore_warnings()
@@ -2686,3 +2685,36 @@ def score(self, X, y):
     grid_search.fit(X, y)
     for param in param_grid:
         assert grid_search.cv_results_[f"param_{param}"].dtype == object
+
+
+def test_search_with_estimators_issue_29157():
+    """Check cv_results_ for estimators with a `dtype` parameter, e.g. OneHotEncoder."""
+    pd = pytest.importorskip("pandas")
+    df = pd.DataFrame(
+        {
+            "numeric_1": [1, 2, 3, 4, 5],
+            "object_1": ["a", "a", "a", "a", "a"],
+            "target": [1.0, 4.1, 2.0, 3.0, 1.0],
+        }
+    )
+    X = df.drop("target", axis=1)
+    y = df["target"]
+    enc = ColumnTransformer(
+        [("enc", OneHotEncoder(sparse_output=False), ["object_1"])],
+        remainder="passthrough",
+    )
+    pipe = Pipeline(
+        [
+            ("enc", enc),
+            ("regressor", LinearRegression()),
+        ]
+    )
+    grid_params = {
+        "enc__enc": [
+            OneHotEncoder(sparse_output=False),
+            OrdinalEncoder(),
+        ]
+    }
+    grid_search = GridSearchCV(pipe, grid_params, cv=2)
+    grid_search.fit(X, y)
+    assert grid_search.cv_results_["param_enc__enc"].dtype == object
