scikit-learn · glemaitre · Jul 27, 2021 · Jun 26, 2021 · Jun 26, 2021 · Jun 26, 2021
diff --git a/doc/whats_new/v1.0.rst b/doc/whats_new/v1.0.rst
@@ -337,6 +337,11 @@ Changelog
   when the variance threshold is negative.
   :pr:`20207` by :user:`Tomohiro Endo <europeanplaice>`
 
+- |Enhancement| :func:`feature_selection.RFE.fit` accepts additional estimator
+  parameters that are passed directly to the estimator's `fit` method.
+  :pr:`20380` by :user:`Iván Pulido <ijpulidos>`, :user:`Felipe Bidu <fbidu>`,
+  :user:`Gil Rutter <g-rutter>`, and :user:`Adrin Jalali <adrinjalali>`.
+
 - |FIX| Fix a bug in :func:`isotonic.isotonic_regression` where the
   `sample_weight` passed by a user were overwritten during the fit.
   :pr:`20515` by :user:`Carsten Allefeld <allefeld>`.

diff --git a/sklearn/feature_selection/_rfe.py b/sklearn/feature_selection/_rfe.py
@@ -192,7 +192,7 @@ def classes_(self):
         """
         return self.estimator_.classes_
 
-    def fit(self, X, y):
+    def fit(self, X, y, **fit_params):
         """Fit the RFE model and then the underlying estimator on the selected features.
 
         Parameters
@@ -203,14 +203,18 @@ def fit(self, X, y):
         y : array-like of shape (n_samples,)
             The target values.
 
+        **fit_params : dict
+            Additional parameters passed to the `fit` method of the underlying
+            estimator.
+
         Returns
         -------
         self : object
             Fitted estimator.
         """
-        return self._fit(X, y)
+        return self._fit(X, y, **fit_params)
 
-    def _fit(self, X, y, step_score=None):
+    def _fit(self, X, y, step_score=None, **fit_params):
         # Parameter step_score controls the calculation of self.scores_
         # step_score is not exposed to users
         # and is used when implementing RFECV
@@ -269,7 +273,7 @@ def _fit(self, X, y, step_score=None):
             if self.verbose > 0:
                 print("Fitting estimator with %d features." % np.sum(support_))
 
-            estimator.fit(X[:, features], y)
+            estimator.fit(X[:, features], y, **fit_params)
 
             # Get importance and rank them
             importances = _get_feature_importances(
@@ -296,7 +300,7 @@ def _fit(self, X, y, step_score=None):
         # Set final attributes
         features = np.arange(n_features)[support_]
         self.estimator_ = clone(self.estimator)
-        self.estimator_.fit(X[:, features], y)
+        self.estimator_.fit(X[:, features], y, **fit_params)
 
         # Compute step score when only n_features_to_select features left
         if step_score:
@@ -325,7 +329,7 @@ def predict(self, X):
         return self.estimator_.predict(self.transform(X))
 
     @if_delegate_has_method(delegate="estimator")
-    def score(self, X, y):
+    def score(self, X, y, **fit_params):
         """Reduce X to the selected features and return the score of the underlying estimator.
 
         Parameters
@@ -336,14 +340,20 @@ def score(self, X, y):
         y : array of shape [n_samples]
             The target values.
 
+        **fit_params : dict
+            Parameters to pass to the `score` method of the underlying
+            estimator.
+
+            .. versionadded:: 1.0
+
         Returns
         -------
         score : float
             Score of the underlying base estimator computed with the selected
             features returned by `rfe.transform(X)` and `y`.
         """
         check_is_fitted(self)
-        return self.estimator_.score(self.transform(X), y)
+        return self.estimator_.score(self.transform(X), y, **fit_params)
 
     def _get_support_mask(self):
         check_is_fitted(self)

diff --git a/sklearn/feature_selection/tests/test_rfe.py b/sklearn/feature_selection/tests/test_rfe.py
@@ -8,6 +8,7 @@
 from numpy.testing import assert_array_almost_equal, assert_array_equal
 from scipy import sparse
 
+from sklearn.base import BaseEstimator, ClassifierMixin
 from sklearn.feature_selection import RFE, RFECV
 from sklearn.datasets import load_iris, make_friedman1
 from sklearn.metrics import zero_one_loss
@@ -108,6 +109,31 @@ def test_rfe():
     assert_array_almost_equal(X_r, X_r_sparse.toarray())
 
 
+def test_RFE_fit_score_params():
+    # Make sure RFE passes the metadata down to fit and score methods of the
+    # underlying estimator
+    class TestEstimator(BaseEstimator, ClassifierMixin):
+        def fit(self, X, y, prop=None):
+            if prop is None:
+                raise ValueError("fit: prop cannot be None")
+            self.svc_ = SVC(kernel="linear").fit(X, y)
+            self.coef_ = self.svc_.coef_
+            return self
+
+        def score(self, X, y, prop=None):
+            if prop is None:
+                raise ValueError("score: prop cannot be None")
+            return self.svc_.score(X, y)
+
+    X, y = load_iris(return_X_y=True)
+    with pytest.raises(ValueError, match="fit: prop cannot be None"):
+        RFE(estimator=TestEstimator()).fit(X, y)
+    with pytest.raises(ValueError, match="score: prop cannot be None"):
+        RFE(estimator=TestEstimator()).fit(X, y, prop="foo").score(X, y)
+
+    RFE(estimator=TestEstimator()).fit(X, y, prop="foo").score(X, y, prop="foo")
+
+
 @pytest.mark.parametrize("n_features_to_select", [-1, 2.1])
 def test_rfe_invalid_n_features_errors(n_features_to_select):
     clf = SVC(kernel="linear")