scikit-learn
diff --git a/‎sklearn/feature_selection/base.py‎
Lines changed: 3 additions & 0 deletions b/‎sklearn/feature_selection/base.py‎
Lines changed: 3 additions & 0 deletions
diff --git a/‎sklearn/feature_selection/tests/test_feature_select.py‎
Lines changed: 18 additions & 1 deletion b/‎sklearn/feature_selection/tests/test_feature_select.py‎
Lines changed: 18 additions & 1 deletion
diff --git a/‎sklearn/feature_selection/univariate_selection.py‎
Lines changed: 4 additions & 2 deletions b/‎sklearn/feature_selection/univariate_selection.py‎
Lines changed: 4 additions & 2 deletions
@@ -73,6 +73,9 @@ def transform(self, X):
         """
         X = check_array(X, accept_sparse='csr')
         mask = self.get_support()
+        if not mask.any():
+            raise ValueError("No features were selected: either the data is"
+                             " too noisy or the selection test too strict.")
         if len(mask) != X.shape[1]:
             raise ValueError("X has a different shape than during fitting.")
         return check_array(X, accept_sparse='csr')[:, safe_mask(X, mask)]
 
@@ -9,6 +9,7 @@
 from sklearn.utils.testing import assert_equal
 from sklearn.utils.testing import assert_almost_equal
 from sklearn.utils.testing import assert_raises
+from sklearn.utils.testing import assert_raise_message
 from sklearn.utils.testing import assert_true
 from sklearn.utils.testing import assert_array_equal
 from sklearn.utils.testing import assert_array_almost_equal
@@ -251,10 +252,12 @@ def test_select_kbest_zero():
                                shuffle=False, random_state=0)
 
     univariate_filter = SelectKBest(f_classif, k=0)
-    univariate_filter.fit(X, y).transform(X)
+    univariate_filter.fit(X, y)
     support = univariate_filter.get_support()
     gtruth = np.zeros(10, dtype=bool)
     assert_array_equal(support, gtruth)
+    assert_raise_message(ValueError, 'No features were selected',
+                         univariate_filter.transform, X)
 
 
 def test_select_fpr_classif():
@@ -585,3 +588,17 @@ def test_f_classif_constant_feature():
     X, y = make_classification(n_samples=10, n_features=5)
     X[:, 0] = 2.0
     assert_warns(UserWarning, f_classif, X, y)
+
+
+def test_no_feature_selected():
+    rng = np.random.RandomState(0)
+
+    # Generate random uncorrelated data: a strict univariate test should
+    # rejects all the features
+    X = rng.rand(40, 10)
+    y = rng.randint(0, 4, size=40)
+    fdr = SelectFdr(alpha=0.00001).fit(X, y)
+    assert_array_equal(fdr.get_support(), np.zeros(10))
+
+    assert_raise_message(ValueError, 'No features were selected',
+                         fdr.transform, X)
@@ -496,8 +496,10 @@ def _get_support_mask(self):
 
         alpha = self.alpha
         sv = np.sort(self.pvalues_)
-        threshold = sv[sv < 
6853
alpha * np.arange(len(self.pvalues_))].max()
-        return self.pvalues_ <= threshold
+        selected = sv[sv < alpha * np.arange(len(self.pvalues_))]
+        if selected.size == 0:
+            return np.zeros_like(self.pvalues_, dtype=bool)
+        return self.pvalues_ <= selected.max()
 
 
 class SelectFwe(_BaseFilter):