FIX array_equal for numpy < 1.8.1

sklearn/metrics/ranking.py

@@ -28,6 +28,7 @@ from ..utils import column_or_1d, check_array
 from ..utils.multiclass import type_of_target
 from ..utils.fixes import isclose
 from ..utils.fixes import bincount
+from ..utils.fixes import array_equal
 from ..utils.stats import rankdata
 from ..utils.sparsefuncs import count_nonzero
 
@@ -295,11 +296,11 @@ def _binary_clf_curve(y_true, y_score, pos_label=None, sample_weight=None):
     # ensure binary classification if pos_label is not specified
     classes = np.unique(y_true)
     if (pos_label is None and
-        not (np.array_equal(classes, [0, 1]) or
-             np.array_equal(classes, [-1, 1]) or
-             np.array_equal(classes, [0]) or
-             np.array_equal(classes, [-1]) or
-             np.array_equal(classes, [1]))):
+        not (array_equal(classes, [0, 1]) or
+             array_equal(classes, [-1, 1]) or
+             array_equal(classes, [0]) or
+             array_equal(classes, [-1]) or
+             array_equal(classes, [1]))):
         raise ValueError("Data is not binary and pos_label is not specified")
     elif pos_label is None:
         pos_label = 1.

sklearn/utils/fixes.py

@@ -375,3 +375,17 @@ else:
             if (not exist_ok or e.errno != errno.EEXIST
                     or not os.path.isdir(name)):
                 raise
+
+
+if np_version < (1, 8, 1):
+    def array_equal(a1, a2):
+        # copy-paste from numpy 1.8.1
+        try:
+            a1, a2 = np.asarray(a1), np.asarray(a2)
+        except:
+            return False
+        if a1.shape != a2.shape:
+            return False
+        return bool(np.asarray(a1 == a2).all())
+else:
+    from numpy import array_equal

sklearn/utils/multiclass.py

@@ -19,10 +19,9 @@ from scipy.sparse import lil_matrix
 import numpy as np
 
 from ..externals.six import string_types
-
 from .validation import check_array
-
 from ..utils.fixes import bincount
+from ..utils.fixes import array_equal
 
 
 def _unique_multiclass(y):
@@ -283,7 +282,7 @@ def _check_partial_fit_first_call(clf, classes=None):
 
     elif classes is not None:
         if getattr(clf, 'classes_', None) is not None:
-            if not np.array_equal(clf.classes_, unique_labels(classes)):
+            if not array_equal(clf.classes_, unique_labels(classes)):
                 raise ValueError(
                     "`classes=%r` is not the same as on last call "
                     "to partial_fit, was: %r" % (classes, clf.classes_))