add support for clf.classes_ to various classifiers

Author: kmike

lightning/impl/adagrad.py

@@ -86,8 +86,8 @@ def _get_loss(self):
         return losses[self.loss]
 
     def fit(self, X, y):
-        self.label_binarizer_ = LabelBinarizer(neg_label=-1, pos_label=1)
-        Y = np.asfortranarray(self.label_binarizer_.fit_transform(y),
+        self._set_label_transformers(y)
+        Y = np.asfortranarray(self.label_binarizer_.transform(y),
                               dtype=np.float64)
         return self._fit(X, Y)
 

lightning/impl/dual_cd.py

@@ -124,8 +124,8 @@ def fit(self, X, y):
         n_samples, n_features = X.shape
         rs = self._get_random_state()
 
-        self.label_binarizer_ = LabelBinarizer(neg_label=-1, pos_label=1)
-        Y = np.asfortranarray(self.label_binarizer_.fit_transform(y),
+        self._set_label_transformers(y)
+        Y = np.asfortranarray(self.label_binarizer_.transform(y),
                               dtype=np.float64)
         n_vectors = Y.shape[1]
 

lightning/impl/sag.py

@@ -160,7 +160,7 @@ def fit(self, X, y, sample_weight=None):
             raise ValueError('Penalties in SAGClassifier. Please use '
                              'SAGAClassifier instead.'
                              '.')
-        self._set_label_transformers(y, neg_label=-1)[0]
+        self._set_label_transformers(y)
         y_binary = self.label_binarizer_.transform(y).astype(np.float64)
         return self._fit(X, y_binary, sample_weight)
 

lightning/impl/sdca.py

@@ -131,8 +131,8 @@ def _get_loss(self):
         return losses[self.loss]
 
     def fit(self, X, y):
-        self.label_binarizer_ = LabelBinarizer(neg_label=-1, pos_label=1)
-        Y = np.asfortranarray(self.label_binarizer_.fit_transform(y),
+        self._set_label_transformers(y)
+        Y = np.asfortranarray(self.label_binarizer_.transform(y),
                               dtype=np.float64)
         return self._fit(X, Y)
 

lightning/impl/svrg.py

@@ -82,8 +82,8 @@ def _get_loss(self):
         return losses[self.loss]
 
     def fit(self, X, y):
-        self.label_binarizer_ = LabelBinarizer(neg_label=-1, pos_label=1)
-        Y = np.asfortranarray(self.label_binarizer_.fit_transform(y),
+        self._set_label_transformers(y)
+        Y = np.asfortranarray(self.label_binarizer_.transform(y),
                               dtype=np.float64)
         return self._fit(X, Y)
 

lightning/impl/tests/test_adagrad.py

@@ -1,7 +1,6 @@
 import numpy as np
 
 from sklearn.datasets import load_iris
-from sklearn.preprocessing import Normalizer
 from sklearn.utils.testing import assert_equal
 from sklearn.utils.testing import assert_almost_equal
 
@@ -42,6 +41,20 @@ def test_adagrad_hinge_multiclass():
     assert_almost_equal(clf.score(X, y), 0.960, 3)
 
 
+def test_adagrad_classes_binary():
+    clf = AdaGradClassifier()
+    assert not hasattr(clf, 'classes_')
+    clf.fit(X_bin, y_bin)
+    assert_equal(list(clf.classes_), [-1, 1])
+
+
+def test_adagrad_classes_multiclass():
+    clf = AdaGradClassifier()
+    assert not hasattr(clf, 'classes_')
+    clf.fit(X, y)
+    assert_equal(list(clf.classes_), [0, 1, 2])
+
+
 def test_adagrad_callback():
     class Callback(object):
 

lightning/impl/tests/test_dual_cd.py

@@ -49,6 +49,7 @@ def test_fit_linear_binary():
         for loss in ("l1", "l2"):
             clf = LinearSVC(loss=loss, random_state=0, max_iter=10)
             clf.fit(data, bin_target)
+            assert_equal(list(clf.classes_), [0, 1])
             assert_equal(clf.score(data, bin_target), 1.0)
             y_pred = clf.decision_function(data).ravel()
 
@@ -66,6 +67,7 @@ def test_fit_linear_multi():
     for data in (mult_dense, mult_sparse):
         clf = LinearSVC(random_state=0)
         clf.fit(data, mult_target)
+        assert_equal(list(clf.classes_), [0, 1, 2])
         y_pred = clf.predict(data)
         acc = np.mean(y_pred == mult_target)
         assert_greater(acc, 0.85)

lightning/impl/tests/test_fista.py

@@ -5,6 +5,7 @@
 
 from sklearn.utils.testing import assert_almost_equal
 from sklearn.utils.testing import assert_true
+from sklearn.utils.testing import assert_equal
 
 from sklearn.datasets import load_digits
 
@@ -107,6 +108,18 @@ def test_fista_multiclass_trace():
         assert_almost_equal(clf.score(data, mult_target), 0.98, 2)
 
 
+def test_fista_bin_classes():
+    clf = FistaClassifier()
+    clf.fit(bin_dense, bin_target)
+    assert_equal(list(clf.classes_), [0, 1])
+
+
+def test_fista_multiclass_classes():
+    clf = FistaClassifier()
+    clf.fit(mult_dense, mult_target)
+    assert_equal(list(clf.classes_), [0, 1, 2])
+
+
 def test_fista_regression():
     reg = FistaRegressor(max_iter=100, verbose=0)
     reg.fit(bin_dense, bin_target)

lightning/impl/tests/test_primal_cd.py

@@ -403,3 +403,15 @@ def test_multiclass_error_nongrouplasso():
     for penalty in ['l1', 'l2']:
         clf = CDClassifier(multiclass=True, penalty=penalty)
         assert_raises(NotImplementedError, clf.fit, mult_dense, mult_target)
+
+
+def test_bin_classes():
+    clf = CDClassifier()
+    clf.fit(bin_dense, bin_target)
+    assert_equal(list(clf.classes_), [0, 1])
+
+
+def test_multiclass_classes():
+    clf = CDClassifier()
+    clf.fit(mult_dense, mult_target)
+    assert_equal(list(clf.classes_), [0, 1, 2])

lightning/impl/tests/test_primal_newton.py

@@ -1,4 +1,5 @@
 from sklearn.utils.testing import assert_almost_equal
+from sklearn.utils.testing import assert_equal
 
 from lightning.impl.datasets.samples_generator import make_classification
 from lightning.impl.primal_newton import KernelSVC
@@ -11,3 +12,4 @@ def test_kernel_svc():
     clf = KernelSVC(kernel="rbf", gamma=0.1, random_state=0, verbose=0)
     clf.fit(bin_dense, bin_target)
     assert_almost_equal(clf.score(bin_dense, bin_target), 1.0)
+    assert_equal(list(clf.classes_), [0, 1])