Review feedback.

scikit-learn · Sep 24, 2019 · 8cef195 · 8cef195
1 parent 2b7dab1
commit 8cef195
Showing 1 changed file with 18 additions and 5 deletions.
diff --git a/sklearn/utils/estimator_checks.py b/sklearn/utils/estimator_checks.py
@@ -1518,7 +1518,10 @@ def check_classifier_multioutput(name, estimator):
     if hasattr(estimator, "decision_function"):
         decision = estimator.decision_function(X)
         assert isinstance(decision, np.ndarray)
-        assert decision.shape == (n_samples, n_labels)
+        assert (decision.shape == (n_samples, n_labels),
+                "The shape of the decision function output  for"
+                " multioutput data is incorrect. Expected {}, got {}."
+                .format((n_samples, n_labels), decision.shape))
 
         dec_pred = (decision > 0).astype(np.int)
         dec_exp = estimator.classes_[dec_pred]
@@ -1529,13 +1532,19 @@ def check_classifier_multioutput(name, estimator):
 
         if isinstance(y_prob, list) and not tags['poor_score']:
             for i in range(n_labels):
-                y_prob[i].shape == (n_samples, n_labels)
+                assert (y_prob[i].shape == (n_samples, n_labels),
+                        "The shape of the probability for multioutput data is"
+                        " incorrect. Expected {}, got {}."
+                        .format((n_samples, n_labels), y_prob[i].shape))
                 assert_array_equal(
                     np.argmax(y_prob[i], axis=1).astype(np.int),
                     y_pred[:, i]
                 )
         elif not tags['poor_score']:
-            y_prob.shape == (n_samples, n_labels)
+            assert (y_prob.shape == (n_samples, n_labels),
+                    "The shape of the probability for multioutput data is"
+                    " incorrect. Expected {}, got {}."
+                    .format((n_samples, n_labels), y_prob.shape))
             assert_array_equal(y_prob.round().astype(int), y_pred)
 
     if (hasattr(estimator, "decision_function") and
@@ -1561,8 +1570,12 @@ def check_regressor_multioutput(name, estimator):
     estimator.fit(X, y)
     y_pred = estimator.predict(X)
 
-    assert y_pred.dtype == np.dtype('float')
-    assert y_pred.shape == y.shape
+    assert (y_pred.dtype == np.dtype('f'),
+            "Multioutput predictions by a regressor are expected to be"
+            " floating-point precision. Got {} instead".format(y_pred.dtype))
+    assert (y_pred.shape == y.shape,
+            "The shape of the orediction for multioutput data is incorrect."
+            " Expected {}, got {}.")
 
 
 @ignore_warnings(category=(DeprecationWarning, FutureWarning))