ENH warn in classification_report when target_names doesn't equal labels size (scikit-learn#7802)

Kenneth Myers · sergeyf · commit 568c998e4831 · 2017-02-28T14:07:15.000-08:00
* Added warning for classification_report when target_names doesn't equal labels size and tests for such a case.
diff --git a/sklearn/metrics/classification.py b/sklearn/metrics/classification.py
@@ -1396,6 +1396,12 @@ class 2       1.00      0.67      0.80         3
     else:
         labels = np.asarray(labels)
 
+    if target_names is not None and len(labels) != len(target_names):
+        warnings.warn(
+            "labels size, {0}, does not match size of target_names, {1}"
+            .format(len(labels), len(target_names))
+        )
+
     last_line_heading = 'avg / total'
 
     if target_names is None:
diff --git a/sklearn/metrics/tests/test_classification.py b/sklearn/metrics/tests/test_classification.py
@@ -722,6 +722,18 @@ def test_classification_report_multiclass_with_long_string_label():
     assert_equal(report, expected_report)
 
 
+def test_classification_report_labels_target_names_unequal_length():
+    y_true = [0, 0, 2, 0, 0]
+    y_pred = [0, 2, 2, 0, 0]
+    target_names = ['class 0', 'class 1', 'class 2']
+
+    assert_warns_message(UserWarning,
+                         "labels size, 2, does not "
+                         "match size of target_names, 3",
+                         classification_report,
+                         y_true, y_pred, target_names=target_names)
+
+
 def test_multilabel_classification_report():
     n_classes = 4
     n_samples = 50
