def test_multilabel_classification():
# Test that multi-label classification works as expected.
# test fit method
X, y = make_multilabel_classification(n_samples=50,
random_state=0,
return_indicator=True)
mlp = MLPClassifier(solver='lbfgs',
hidden_layer_sizes=50,
alpha=1e-5,
max_iter=150,
random_state=0,
activation='logistic',
learning_rate_init=0.2)
mlp.fit(X, y)
assert mlp.score(X, y) > 0.97
# test partial fit method
mlp = MLPClassifier(solver='sgd',
hidden_layer_sizes=50,
max_iter=150,
random_state=0,
activation='logistic',
alpha=1e-5,
learning_rate_init=0.2)
for i in range(100):
mlp.partial_fit(X, y, classes=[0, 1, 2, 3, 4])
assert mlp.score(X, y) > 0.9
# Make sure early stopping still work now that spliting is stratified by
# default (it is disabled for multilabel classification)
mlp = MLPClassifier(early_stopping=True)
mlp.fit(X, y).predict(X)
def test_partial_fit_classification():
# Test partial_fit on classification.
# `partial_fit` should yield the same results as 'fit' for binary and
# multi-class classification.
for X, y in classification_datasets:
X = X
y = y
mlp = MLPClassifier(solver='sgd',
max_iter=100,
random_state=1,
tol=0,
alpha=1e-5,
learning_rate_init=0.2)
with ignore_warnings(category=ConvergenceWarning):
mlp.fit(X, y)
pred1 = mlp.predict(X)
mlp = MLPClassifier(solver='sgd',
random_state=1,
alpha=1e-5,
learning_rate_init=0.2)
for i in range(100):
mlp.partial_fit(X, y, classes=np.unique(y))
pred2 = mlp.predict(X)
assert_array_equal(pred1, pred2)
assert mlp.score(X, y) > 0.95
def plot_on_dataset(X, y, ax, name):
# for each dataset, plot learning for each learning strategy
print("\nlearning on dataset %s" % name)
ax.set_title(name)
X = MinMaxScaler().fit_transform(X)
mlps = []
if name == "digits":
# digits is larger but converges fairly quickly
max_iter = 15
else:
max_iter = 400
for label, param in zip(labels, params):
print("training: %s" % label)
mlp = MLPClassifier(random_state=0, max_iter=max_iter, **param)
# some parameter combinations will not converge as can be seen on the
# plots so they are ignored here
with warnings.catch_warnings():
warnings.filterwarnings("ignore",
category=ConvergenceWarning,
module="mrex")
mlp.fit(X, y)
mlps.append(mlp)
print("Training set score: %f" % mlp.score(X, y))
print("Training set loss: %f" % mlp.loss_)
for mlp, label, args in zip(mlps, labels, plot_args):
ax.plot(mlp.loss_curve_, label=label, **args)
def test_partial_fit_unseen_classes():
# Non regression test for bug 6994
# Tests for labeling errors in partial fit
clf = MLPClassifier(random_state=0)
clf.partial_fit([[1], [2], [3]], ["a", "b", "c"],
classes=["a", "b", "c", "d"])
clf.partial_fit([[4]], ["d"])
assert clf.score([[1], [2], [3], [4]], ["a", "b", "c", "d"]) > 0
def test_lbfgs_classification(X, y):
# Test lbfgs on classification.
# It should achieve a score higher than 0.95 for the binary and multi-class
# versions of the digits dataset.
X_train = X[:150]
y_train = y[:150]
X_test = X[150:]
expected_shape_dtype = (X_test.shape[0], y_train.dtype.kind)
for activation in ACTIVATION_TYPES:
mlp = MLPClassifier(solver='lbfgs',
hidden_layer_sizes=50,
max_iter=150,
shuffle=True,
random_state=1,
activation=activation)
mlp.fit(X_train, y_train)
y_predict = mlp.predict(X_test)
assert mlp.score(X_train, y_train) > 0.95
assert ((y_predict.shape[0],
y_predict.dtype.kind) == expected_shape_dtype)
X = X / 255.
# rescale the data, use the traditional train/test split
X_train, X_test = X[:60000], X[60000:]
y_train, y_test = y[:60000], y[60000:]
mlp = MLPClassifier(hidden_layer_sizes=(50, ),
max_iter=10,
alpha=1e-4,
solver='sgd',
verbose=10,
random_state=1,
learning_rate_init=.1)
mlp.fit(X_train, y_train)
print("Training set score: %f" % mlp.score(X_train, y_train))
print("Test set score: %f" % mlp.score(X_test, y_test))
fig, axes = plt.subplots(4, 4)
# use global min / max to ensure all weights are shown on the same scale
vmin, vmax = mlp.coefs_[0].min(), mlp.coefs_[0].max()
for coef, ax in zip(mlp.coefs_[0].T, axes.ravel()):
ax.matshow(coef.reshape(28, 28),
cmap=plt.cm.gray,
vmin=.5 * vmin,
vmax=.5 * vmax)
ax.set_xticks(())
ax.set_yticks(())
plt.show()