def test_predict_proba_multiclass():
# Test that predict_proba works as expected for multi class.
X = X_digits_multi[:10]
y = y_digits_multi[:10]
clf = MLPClassifier(hidden_layer_sizes=5)
with ignore_warnings(category=ConvergenceWarning):
clf.fit(X, y)
y_proba = clf.predict_proba(X)
y_log_proba = clf.predict_log_proba(X)
(n_samples, n_classes) = y.shape[0], np.unique(y).size
proba_max = y_proba.argmax(axis=1)
proba_log_max = y_log_proba.argmax(axis=1)
assert y_proba.shape == (n_samples, n_classes)
assert_array_equal(proba_max, proba_log_max)
assert_array_equal(y_log_proba, np.log(y_proba))
def test_predict_proba_multilabel():
# Test that predict_proba works as expected for multilabel.
# Multilabel should not use softmax which makes probabilities sum to 1
X, Y = make_multilabel_classification(n_samples=50,
random_state=0,
return_indicator=True)
n_samples, n_classes = Y.shape
clf = MLPClassifier(solver='lbfgs', hidden_layer_sizes=30, random_state=0)
clf.fit(X, Y)
y_proba = clf.predict_proba(X)
assert y_proba.shape == (n_samples, n_classes)
assert_array_equal(y_proba > 0.5, Y)
y_log_proba = clf.predict_log_proba(X)
proba_max = y_proba.argmax(axis=1)
proba_log_max = y_log_proba.argmax(axis=1)
assert (y_proba.sum(1) - 1).dot(y_proba.sum(1) - 1) > 1e-10
assert_array_equal(proba_max, proba_log_max)
assert_array_equal(y_log_proba, np.log(y_proba))
def test_predict_proba_binary():
# Test that predict_proba works as expected for binary class.
X = X_digits_binary[:50]
y = y_digits_binary[:50]
clf = MLPClassifier(hidden_layer_sizes=5,
activation='logistic',
random_state=1)
with ignore_warnings(category=ConvergenceWarning):
clf.fit(X, y)
y_proba = clf.predict_proba(X)
y_log_proba = clf.predict_log_proba(X)
(n_samples, n_classes) = y.shape[0], 2
proba_max = y_proba.argmax(axis=1)
proba_log_max = y_log_proba.argmax(axis=1)
assert y_proba.shape == (n_samples, n_classes)
assert_array_equal(proba_max, proba_log_max)
assert_array_equal(y_log_proba, np.log(y_proba))
assert roc_auc_score(y, y_proba[:, 1]) == 1.0