def test_split_classifier_with_single_classifier(self):
c = sklearn.dummy.DummyClassifier('constant', constant=0)
sc = SplitClassifier(c, lambda X: np.arange(len(X)) % 2)
X = np.ones(shape=(100, 3))
y = np.zeros(100)
X_test = np.ones(shape=(100, 3))
sc.fit(X, y)
sc.classifiers[1].constant = 1
predictions = sc.predict(X_test)
self.assertEquals((100,), predictions.shape)
np.testing.assert_array_equal(np.arange(100) % 2, predictions)
def test_split_classifier_with_3_indexes(self):
c0 = sklearn.dummy.DummyClassifier('constant', constant=0)
c1 = sklearn.dummy.DummyClassifier('constant', constant=1)
c2 = sklearn.dummy.DummyClassifier('constant', constant=2)
sc = SplitClassifier((c0, c1, c2), lambda X: np.arange(len(X)) % 3)
X = np.ones(shape=(100, 3))
y = np.arange(100) % 3
X_test = np.ones(shape=(100, 3))
sc.fit(X, y)
predictions = sc.predict(X_test)
self.assertEquals((100,), predictions.shape)
np.testing.assert_array_equal(np.arange(100) % 3, predictions)
def test_split_classifier_with_single_classifier(self):
c = sklearn.dummy.DummyClassifier('constant', constant=0)
sc = SplitClassifier(c, lambda X: np.arange(len(X)) % 2)
X = np.ones(shape=(100, 3))
y = np.zeros(100)
X_test = np.ones(shape=(100, 3))
sc.fit(X, y)
sc.classifiers[1].constant = 1
predictions = sc.predict(X_test)
self.assertEquals((100, ), predictions.shape)
np.testing.assert_array_equal(np.arange(100) % 2, predictions)
def test_split_classifier_with_3_indexes(self):
c0 = sklearn.dummy.DummyClassifier('constant', constant=0)
c1 = sklearn.dummy.DummyClassifier('constant', constant=1)
c2 = sklearn.dummy.DummyClassifier('constant', constant=2)
sc = SplitClassifier((c0, c1, c2), lambda X: np.arange(len(X)) % 3)
X = np.ones(shape=(100, 3))
y = np.arange(100) % 3
X_test = np.ones(shape=(100, 3))
sc.fit(X, y)
predictions = sc.predict(X_test)
self.assertEquals((100, ), predictions.shape)
np.testing.assert_array_equal(np.arange(100) % 3, predictions)
def test_split_classifier_with_fallback_classifier(self):
c0 = sklearn.dummy.DummyClassifier('constant', constant=0)
c1 = sklearn.dummy.DummyClassifier('constant', constant=1)
fallback = sklearn.dummy.DummyClassifier('constant', constant=999)
def indexer(X):
indexes = np.arange(len(X)) % 3
indexes[indexes==2] = -1
return indexes
sc = SplitClassifier((c0, c1), indexer, fallback_classifier=fallback)
X = np.ones(shape=(100, 3))
y = np.arange(100) % 3
X_test = np.ones(shape=(100, 3))
fallback.fit(X, np.array([999] * 100))
sc.fit(X, y)
predictions = sc.predict(X_test)
self.assertEquals((100,), predictions.shape)
expected = np.arange(100) % 3
expected[expected==2] = 999
np.testing.assert_array_equal(expected, predictions)
def test_split_classifier_with_fallback_classifier(self):
c0 = sklearn.dummy.DummyClassifier('constant', constant=0)
c1 = sklearn.dummy.DummyClassifier('constant', constant=1)
fallback = sklearn.dummy.DummyClassifier('constant', constant=999)
def indexer(X):
indexes = np.arange(len(X)) % 3
indexes[indexes == 2] = -1
return indexes
sc = SplitClassifier((c0, c1), indexer, fallback_classifier=fallback)
X = np.ones(shape=(100, 3))
y = np.arange(100) % 3
X_test = np.ones(shape=(100, 3))
fallback.fit(X, np.array([999] * 100))
sc.fit(X, y)
predictions = sc.predict(X_test)
self.assertEquals((100, ), predictions.shape)
expected = np.arange(100) % 3
expected[expected == 2] = 999
np.testing.assert_array_equal(expected, predictions)
