BUG fix behaviour in confusion_matrix with with empty array-like as input

doc/whats_new/v0.23.rst

@@ -202,6 +202,10 @@ Changelog
 - |Fix| Fixed a bug in :func:`metrics.mutual_info_score` where negative
   scores could be returned. :pr:`16362` by `Thomas Fan`_.
 
+- |Fix| Fixed a bug in :func:`metrics.confusion_matrix` that would raise
+  an error when `y_true` and `y_pred` were length zero and `labels` was
+  not `None`. :pr:`16442` by `Kyle Parsons <parsons-kyle-89>`
+
 :mod:`sklearn.model_selection`
 ..............................
 

sklearn/metrics/_classification.py

@@ -274,6 +274,9 @@ def confusion_matrix(y_true, y_pred, labels=None, sample_weight=None,
 
     if labels is None:
         labels = unique_labels(y_true, y_pred)
+    elif y_true.size == 0:
+        n_labels = len(labels)
+        return np.zeros((n_labels, n_labels), dtype=np.int)
     else:
         labels = np.asarray(labels)
         if np.all([l not in y_true for l in labels]):

sklearn/metrics/_plot/tests/test_plot_confusion_matrix.py

@@ -225,7 +225,14 @@ def test_confusion_matrix_contrast(pyplot):
     assert_allclose(disp.text_[1, 0].get_color(), max_color)
     assert_allclose(disp.text_[1, 1].get_color(), min_color)
 
+    # Non-regression test for #16442
+    cm = np.array([[0, 0], [0, 0]])
+    disp = ConfusionMatrixDisplay(cm, display_labels=[0, 1])
 
+    disp.plot(cmap=pyplot.cm.Blues)
+    min_color = pyplot.cm.Blues(0)
+    for text in disp.text_.ravel():
+        assert_allclose(text.get_color(), min_color)
 
 
 @pytest.mark.parametrize(

sklearn/metrics/tests/test_classification.py

@@ -914,6 +914,18 @@ def test_confusion_matrix_multiclass_subset_labels():
                          labels=[extra_label, extra_label + 1])
 
 
+@pytest.mark.parametrize(
+    'labels', (None, [0, 1], [0, 1, 2]),
+    ids=['None', 'binary', 'multiclass']
+)
+def test_confusion_matrix_on_zero_length_input(labels):
+    labels = [] if not labels else labels
+    n_classes = len(labels)
+    expected = np.zeros((n_classes, n_classes), dtype=np.int)
+    cm = confusion_matrix([], [], labels)
+    assert_array_equal(cm, expected)
+
+
 def test_confusion_matrix_dtype():
     y = [0, 1, 1]
     weight = np.ones(len(y))