ENH allow SelectKBest to select all features in a parameter search

jnothman · larsmans · commit b80f6a80b020 · 2013-03-13T12:28:49.000+01:00
diff --git a/sklearn/feature_selection/tests/test_feature_select.py b/sklearn/feature_selection/tests/test_feature_select.py
@@ -189,6 +189,18 @@ def test_select_kbest_classif():
     assert_array_equal(support, gtruth)
 
 
+def test_select_kbest_all():
+    """
+    Test whether k="all" correctly returns all features.
+    """
+    X, y = make_classification(n_samples=20, n_features=10,
+                               shuffle=False, random_state=0)
+
+    univariate_filter = SelectKBest(f_classif, k='all')
+    X_r = univariate_filter.fit(X, y).transform(X)
+    assert_array_equal(X, X_r)
+
+
 def test_select_fpr_classif():
     """
     Test whether the relative univariate feature selection
diff --git a/sklearn/feature_selection/univariate_selection.py b/sklearn/feature_selection/univariate_selection.py
@@ -401,8 +401,9 @@ class SelectKBest(_ScoreFilter):
         Function taking two arrays X and y, and returning a pair of arrays
         (scores, pvalues).
 
-    k : int, optional, default=10
+    k : int or "all", optional, default=10
         Number of top features to select.
+        The "all" option bypasses selection, for use in a parameter search.
 
     Attributes
     ----------
@@ -425,8 +426,11 @@ def __init__(self, score_func=f_classif, k=10):
 
     def _get_support_mask(self):
         k = self.k
+        if k == 'all':
+            return np.ones(self.scores_.shape, dtype=bool)
         if k > len(self.scores_):
-            raise ValueError("cannot select %d features among %d"
+            raise ValueError("Cannot select %d features among %d. "
+                             "Use k='all' to return all features."
                              % (k, len(self.scores_)))
 
         scores = _clean_nans(self.scores_)
