Remove _make_masks and use OneHotEncoder instead

wdevazelhes · wdevazelhes · Oct 27, 2017 · Oct 27, 2017 · Oct 27, 2017 · Oct 30, 2017
commit 0cee29d78fecc0bf29b60f1008017f75c399a9d5
diff --git a/sklearn/neighbors/nca.py b/sklearn/neighbors/nca.py
@@ -529,24 +529,3 @@ def _check_scalar(x, name, target_type, min_val=None, max_val=None):
 
     if max_val is not None and x > max_val:
         raise ValueError('`{}`= {}, must be <= {}.'.format(name, x, max_val))
-
-
-def _make_masks(y):
-    """Create one-hot encoding of vector ``y``.
-
-    Parameters
-    ----------
-    y : array, shape (n_samples,)
-        Data samples labels.
-
-    Returns
-    -------
-    masks: array, shape (n_samples, n_classes)
-        One-hot encoding of ``y``.
-    """
-    masks = OneHotEncoder(sparse=False, dtype=bool).fit_transform(y[:,
-                                                                  np.newaxis])
-    # n = y.shape[0]
-    # masks = np.zeros((n, y.max() + 1), dtype=bool)
-    # masks[np.arange(n), y] = [True]
-    return masks
diff --git a/sklearn/neighbors/tests/test_nca.py b/sklearn/neighbors/tests/test_nca.py
@@ -1,10 +1,11 @@
 import numpy as np
 from numpy.testing import assert_array_equal
+from sklearn.preprocessing import OneHotEncoder
 from sklearn.utils import check_random_state
 from sklearn.utils.testing import assert_raises, assert_equal
 from sklearn.datasets import load_iris, make_classification
 from sklearn.model_selection import train_test_split
-from sklearn.neighbors.nca import NeighborhoodComponentsAnalysis, _make_masks
+from sklearn.neighbors.nca import NeighborhoodComponentsAnalysis
 from sklearn.metrics import pairwise_distances
 
 
@@ -59,7 +60,8 @@ def test_finite_differences():
     nca = NeighborhoodComponentsAnalysis(None, init=point)
 
     X, y, init = nca._validate_params(X, y)
-    masks = _make_masks(y)
+    masks = OneHotEncoder(sparse=False,
+                          dtype=bool).fit_transform(y[:, np.newaxis])
     diffs = X[:, np.newaxis] - X[np.newaxis]
     nca.n_iter_ = 0