def check_transformer_pickle(name, Transformer):
X, y = make_blobs(n_samples=30, centers=[[0, 0, 0], [1, 1, 1]],
random_state=0, n_features=2, cluster_std=0.1)
n_samples, n_features = X.shape
X = StandardScaler().fit_transform(X)
X -= X.min()
# catch deprecation warnings
with warnings.catch_warnings(record=True):
transformer = Transformer()
if not hasattr(transformer, 'transform'):
return
set_random_state(transformer)
set_fast_parameters(transformer)
# fit
if name in CROSS_DECOMPOSITION:
random_state = np.random.RandomState(seed=12345)
y_ = np.vstack([y, 2 * y + random_state.randint(2, size=len(y))])
y_ = y_.T
else:
y_ = y
transformer.fit(X, y_)
X_pred = transformer.fit(X, y_).transform(X)
pickled_transformer = pickle.dumps(transformer)
unpickled_transformer = pickle.loads(pickled_transformer)
pickled_X_pred = unpickled_transformer.transform(X)
assert_array_almost_equal(pickled_X_pred, X_pred)
def test_feature_union():
# basic sanity check for feature union
iris = load_iris()
X = iris.data
X -= X.mean(axis=0)
y = iris.target
svd = TruncatedSVD(n_components=2, random_state=0)
select = SelectKBest(k=1)
fs = FeatureUnion([("svd", svd), ("select", select)])
fs.fit(X, y)
X_transformed = fs.transform(X)
assert_equal(X_transformed.shape, (X.shape[0], 3))
# check if it does the expected thing
assert_array_almost_equal(X_transformed[:, :-1], svd.fit_transform(X))
assert_array_equal(X_transformed[:, -1],
select.fit_transform(X, y).ravel())
# test if it also works for sparse input
# We use a different svd object to control the random_state stream
fs = FeatureUnion([("svd", svd), ("select", select)])
X_sp = sparse.csr_matrix(X)
X_sp_transformed = fs.fit_transform(X_sp, y)
assert_array_almost_equal(X_transformed, X_sp_transformed.toarray())
# test setting parameters
fs.set_params(select__k=2)
assert_equal(fs.fit_transform(X, y).shape, (X.shape[0], 4))
# test it works with transformers missing fit_transform
fs = FeatureUnion([("mock", TransfT()), ("svd", svd), ("select", select)])
X_transformed = fs.fit_transform(X, y)
assert_equal(X_transformed.shape, (X.shape[0], 8))
def test_RadiusNeighborsRegressor_multioutput_with_uniform_weight():
"""Test radius neighbors in multi-output regression (uniform weight)"""
rng = check_random_state(0)
n_features = 5
n_samples = 40
n_output = 4
X = rng.rand(n_samples, n_features)
y = rng.rand(n_samples, n_output)
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)
for algorithm, weights in product(ALGORITHMS, [None, 'uniform']):
rnn = neighbors. RadiusNeighborsRegressor(weights=weights,
algorithm=algorithm)
rnn.fit(X_train, y_train)
neigh_idx = rnn.radius_neighbors(X_test, return_distance=False)
y_pred_idx = np.array([np.mean(y_train[idx], axis=0)
for idx in neigh_idx])
y_pred_idx = np.array(y_pred_idx)
y_pred = rnn.predict(X_test)
assert_equal(y_pred_idx.shape, y_test.shape)
assert_equal(y_pred.shape, y_test.shape)
assert_array_almost_equal(y_pred, y_pred_idx)
def test_kneighbors_classifier_predict_proba():
"""Test KNeighborsClassifier.predict_proba() method"""
X = np.array([[0, 2, 0],
[0, 2, 1],
[2, 0, 0],
[2, 2, 0],
[0, 0, 2],
[0, 0, 1]])
y = np.array([4, 4, 5, 5, 1, 1])
cls = neighbors.KNeighborsClassifier(n_neighbors=3, p=1) # cityblock dist
cls.fit(X, y)
y_prob = cls.predict_proba(X)
real_prob = np.array([[0, 2. / 3, 1. / 3],
[1. / 3, 2. / 3, 0],
[1. / 3, 0, 2. / 3],
[0, 1. / 3, 2. / 3],
[2. / 3, 1. / 3, 0],
[2. / 3, 1. / 3, 0]])
assert_array_equal(real_prob, y_prob)
# Check that it also works with non integer labels
cls.fit(X, y.astype(str))
y_prob = cls.predict_proba(X)
assert_array_equal(real_prob, y_prob)
# Check that it works with weights='distance'
cls = neighbors.KNeighborsClassifier(
n_neighbors=2, p=1, weights='distance')
cls.fit(X, y)
y_prob = cls.predict_proba(np.array([[0, 2, 0], [2, 2, 2]]))
real_prob = np.array([[0, 1, 0], [0, 0.4, 0.6]])
assert_array_almost_equal(real_prob, y_prob)
def test_feature_union_weights():
# test feature union with transformer weights
iris = load_iris()
X = iris.data
y = iris.target
pca = RandomizedPCA(n_components=2, random_state=0)
select = SelectKBest(k=1)
# test using fit followed by transform
fs = FeatureUnion([("pca", pca), ("select", select)],
transformer_weights={"pca": 10})
fs.fit(X, y)
X_transformed = fs.transform(X)
# test using fit_transform
fs = FeatureUnion([("pca", pca), ("select", select)],
transformer_weights={"pca": 10})
X_fit_transformed = fs.fit_transform(X, y)
# test it works with transformers missing fit_transform
fs = FeatureUnion([("mock", TransfT()), ("pca", pca), ("select", select)],
transformer_weights={"mock": 10})
X_fit_transformed_wo_method = fs.fit_transform(X, y)
# check against expected result
# We use a different pca object to control the random_state stream
assert_array_almost_equal(X_transformed[:, :-1], 10 * pca.fit_transform(X))
assert_array_equal(X_transformed[:, -1],
select.fit_transform(X, y).ravel())
assert_array_almost_equal(X_fit_transformed[:, :-1],
10 * pca.fit_transform(X))
assert_array_equal(X_fit_transformed[:, -1],
select.fit_transform(X, y).ravel())
assert_equal(X_fit_transformed_wo_method.shape, (X.shape[0], 7))
def test_RadiusNeighborsClassifier_multioutput():
"""Test k-NN classifier on multioutput data"""
rng = check_random_state(0)
n_features = 2
n_samples = 40
n_output = 3
X = rng.rand(n_samples, n_features)
y = rng.randint(0, 3, (n_samples, n_output))
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)
weights = [None, 'uniform', 'distance', _weight_func]
for algorithm, weights in product(ALGORITHMS, weights):
# Stack single output prediction
y_pred_so = []
for o in range(n_output):
rnn = neighbors.RadiusNeighborsClassifier(weights=weights,
algorithm=algorithm)
rnn.fit(X_train, y_train[:, o])
y_pred_so.append(rnn.predict(X_test))
y_pred_so = np.vstack(y_pred_so).T
assert_equal(y_pred_so.shape, y_test.shape)
# Multioutput prediction
rnn_mo = neighbors.RadiusNeighborsClassifier(weights=weights,
algorithm=algorithm)
rnn_mo.fit(X_train, y_train)
y_pred_mo = rnn_mo.predict(X_test)
assert_equal(y_pred_mo.shape, y_test.shape)
assert_array_almost_equal(y_pred_mo, y_pred_so)
def test_sparse_design_with_sample_weights():
# Sample weights must work with sparse matrices
n_sampless = [2, 3]
n_featuress = [3, 2]
rng = np.random.RandomState(42)
sparse_matrix_converters = [sp.coo_matrix,
sp.csr_matrix,
sp.csc_matrix,
sp.lil_matrix,
sp.dok_matrix
]
sparse_ridge = Ridge(alpha=1., fit_intercept=False)
dense_ridge = Ridge(alpha=1., fit_intercept=False)
for n_samples, n_features in zip(n_sampless, n_featuress):
X = rng.randn(n_samples, n_features)
y = rng.randn(n_samples)
sample_weights = rng.randn(n_samples) ** 2 + 1
for sparse_converter in sparse_matrix_converters:
X_sparse = sparse_converter(X)
sparse_ridge.fit(X_sparse, y, sample_weight=sample_weights)
dense_ridge.fit(X, y, sample_weight=sample_weights)
assert_array_almost_equal(sparse_ridge.coef_, dense_ridge.coef_,
decimal=6)
def test_primal_dual_relationship():
y = y_diabetes.reshape(-1, 1)
coef = _solve_cholesky(X_diabetes, y, alpha=[1e-2])
K = np.dot(X_diabetes, X_diabetes.T)
dual_coef = _solve_cholesky_kernel(K, y, alpha=[1e-2])
coef2 = np.dot(X_diabetes.T, dual_coef).T
assert_array_almost_equal(coef, coef2)
def test_labels_assignment_and_inertia():
# pure numpy implementation as easily auditable reference gold
# implementation
rng = np.random.RandomState(42)
noisy_centers = centers + rng.normal(size=centers.shape)
labels_gold = - np.ones(n_samples, dtype=np.int)
mindist = np.empty(n_samples)
mindist.fill(np.infty)
for center_id in range(n_clusters):
dist = np.sum((X - noisy_centers[center_id]) ** 2, axis=1)
labels_gold[dist < mindist] = center_id
mindist = np.minimum(dist, mindist)
inertia_gold = mindist.sum()
assert_true((mindist >= 0.0).all())
assert_true((labels_gold != -1).all())
# perform label assignment using the dense array input
x_squared_norms = (X ** 2).sum(axis=1)
labels_array, inertia_array = _labels_inertia(
X, x_squared_norms, noisy_centers)
assert_array_almost_equal(inertia_array, inertia_gold)
assert_array_equal(labels_array, labels_gold)
# perform label assignment using the sparse CSR input
x_squared_norms_from_csr = row_norms(X_csr, squared=True)
labels_csr, inertia_csr = _labels_inertia(
X_csr, x_squared_norms_from_csr, noisy_centers)
assert_array_almost_equal(inertia_csr, inertia_gold)
assert_array_equal(labels_csr, labels_gold)
def test_class_weights():
# Test class weights.
X = np.array([[-1.0, -1.0], [-1.0, 0], [-.8, -1.0],
[1.0, 1.0], [1.0, 0.0]])
y = [1, 1, 1, -1, -1]
clf = RidgeClassifier(class_weight=None)
clf.fit(X, y)
assert_array_equal(clf.predict([[0.2, -1.0]]), np.array([1]))
# we give a small weights to class 1
clf = RidgeClassifier(class_weight={1: 0.001})
clf.fit(X, y)
# now the hyperplane should rotate clock-wise and
# the prediction on this point should shift
assert_array_equal(clf.predict([[0.2, -1.0]]), np.array([-1]))
# check if class_weight = 'balanced' can handle negative labels.
clf = RidgeClassifier(class_weight='balanced')
clf.fit(X, y)
assert_array_equal(clf.predict([[0.2, -1.0]]), np.array([1]))
# class_weight = 'balanced', and class_weight = None should return
# same values when y has equal number of all labels
X = np.array([[-1.0, -1.0], [-1.0, 0], [-.8, -1.0], [1.0, 1.0]])
y = [1, 1, -1, -1]
clf = RidgeClassifier(class_weight=None)
clf.fit(X, y)
clfa = RidgeClassifier(class_weight='balanced')
clfa.fit(X, y)
assert_equal(len(clfa.classes_), 2)
assert_array_almost_equal(clf.coef_, clfa.coef_)
assert_array_almost_equal(clf.intercept_, clfa.intercept_)
def test_cross_val_predict_with_method():
iris = load_iris()
X, y = iris.data, iris.target
X, y = shuffle(X, y, random_state=0)
classes = len(set(y))
kfold = KFold(len(iris.target))
methods = ['decision_function', 'predict_proba', 'predict_log_proba']
for method in methods:
est = LogisticRegression()
predictions = cross_val_predict(est, X, y, method=method)
assert_equal(len(predictions), len(y))
expected_predictions = np.zeros([len(y), classes])
func = getattr(est, method)
# Naive loop (should be same as cross_val_predict):
for train, test in kfold.split(X, y):
est.fit(X[train], y[train])
expected_predictions[test] = func(X[test])
predictions = cross_val_predict(est, X, y, method=method,
cv=kfold)
assert_array_almost_equal(expected_predictions, predictions)
def test_enet_l1_ratio():
# Test that an error message is raised if an estimator that
# uses _alpha_grid is called with l1_ratio=0
msg = ("Automatic alpha grid generation is not supported for l1_ratio=0. "
"Please supply a grid by providing your estimator with the "
"appropriate `alphas=` argument.")
X = np.array([[1, 2, 4, 5, 8], [3, 5, 7, 7, 8]]).T
y = np.array([12, 10, 11, 21, 5])
assert_raise_message(ValueError, msg, ElasticNetCV(
l1_ratio=0, random_state=42).fit, X, y)
assert_raise_message(ValueError, msg, MultiTaskElasticNetCV(
l1_ratio=0, random_state=42).fit, X, y[:, None])
# Test that l1_ratio=0 is allowed if we supply a grid manually
alphas = [0.1, 10]
estkwds = {'alphas': alphas, 'random_state': 42}
est_desired = ElasticNetCV(l1_ratio=0.00001, **estkwds)
est = ElasticNetCV(l1_ratio=0, **estkwds)
with ignore_warnings():
est_desired.fit(X, y)
est.fit(X, y)
assert_array_almost_equal(est.coef_, est_desired.coef_, decimal=5)
est_desired = MultiTaskElasticNetCV(l1_ratio=0.00001, **estkwds)
est = MultiTaskElasticNetCV(l1_ratio=0, **estkwds)
with ignore_warnings():
est.fit(X, y[:, None])
est_desired.fit(X, y[:, None])
assert_array_almost_equal(est.coef_, est_desired.coef_, decimal=5)
def test_skewed_chi2_sampler():
"""test that RBFSampler approximates kernel on random data"""
# compute exact kernel
c = 0.03
# appreviations for easier formular
X_c = (X + c)[:, np.newaxis, :]
Y_c = (Y + c)[np.newaxis, :, :]
# we do it in log-space in the hope that it's more stable
# this array is n_samples_x x n_samples_y big x n_features
log_kernel = ((np.log(X_c) / 2.) + (np.log(Y_c) / 2.) + np.log(2.) -
np.log(X_c + Y_c))
# reduce to n_samples_x x n_samples_y by summing over features in log-space
kernel = np.exp(log_kernel.sum(axis=2))
# approximate kernel mapping
transform = SkewedChi2Sampler(skewedness=c, n_components=1000,
random_state=42)
X_trans = transform.fit_transform(X)
Y_trans = transform.transform(Y)
kernel_approx = np.dot(X_trans, Y_trans.T)
assert_array_almost_equal(kernel, kernel_approx, 1)
# test error is raised on negative input
Y_neg = Y.copy()
Y_neg[0, 0] = -1
assert_raises(ValueError, transform.transform, Y_neg)
def test_inverse_transform():
# Test FastICA.inverse_transform
n_features = 10
n_samples = 100
n1, n2 = 5, 10
rng = np.random.RandomState(0)
X = rng.random_sample((n_samples, n_features))
expected = {(True, n1): (n_features, n1),
(True, n2): (n_features, n2),
(False, n1): (n_features, n2),
(False, n2): (n_features, n2)}
for whiten in [True, False]:
for n_components in [n1, n2]:
n_components_ = (n_components if n_components is not None else
X.shape[1])
ica = FastICA(n_components=n_components, random_state=rng,
whiten=whiten)
with warnings.catch_warnings(record=True):
# catch "n_components ignored" warning
Xt = ica.fit_transform(X)
expected_shape = expected[(whiten, n_components_)]
assert_equal(ica.mixing_.shape, expected_shape)
X2 = ica.inverse_transform(Xt)
assert_equal(X.shape, X2.shape)
# reversibility test in non-reduction case
if n_components == X.shape[1]:
assert_array_almost_equal(X, X2)
def check_regressors_pickle(name, Regressor):
X, y = _boston_subset()
y = StandardScaler().fit_transform(y) # X is already scaled
y = multioutput_estimator_convert_y_2d(name, y)
rnd = np.random.RandomState(0)
# catch deprecation warnings
with warnings.catch_warnings(record=True):
regressor = Regressor()
set_fast_parameters(regressor)
if not hasattr(regressor, 'alphas') and hasattr(regressor, 'alpha'):
# linear regressors need to set alpha, but not generalized CV ones
regressor.alpha = 0.01
if name in CROSS_DECOMPOSITION:
y_ = np.vstack([y, 2 * y + rnd.randint(2, size=len(y))])
y_ = y_.T
else:
y_ = y
regressor.fit(X, y_)
y_pred = regressor.predict(X)
# store old predictions
pickled_regressor = pickle.dumps(regressor)
unpickled_regressor = pickle.loads(pickled_regressor)
pickled_y_pred = unpickled_regressor.predict(X)
assert_array_almost_equal(pickled_y_pred, y_pred)
def check_class_weight_auto_linear_classifier(name, Classifier):
"""Test class weights with non-contiguous class labels."""
X = np.array([[-1.0, -1.0], [-1.0, 0], [-.8, -1.0],
[1.0, 1.0], [1.0, 0.0]])
y = [1, 1, 1, -1, -1]
with warnings.catch_warnings(record=True):
classifier = Classifier()
if hasattr(classifier, "n_iter"):
# This is a very small dataset, default n_iter are likely to prevent
# convergence
classifier.set_params(n_iter=1000)
set_random_state(classifier)
# Let the model compute the class frequencies
classifier.set_params(class_weight='auto')
coef_auto = classifier.fit(X, y).coef_.copy()
# Count each label occurrence to reweight manually
mean_weight = (1. / 3 + 1. / 2) / 2
class_weight = {
1: 1. / 3 / mean_weight,
-1: 1. / 2 / mean_weight,
}
classifier.set_params(class_weight=class_weight)
coef_manual = classifier.fit(X, y).coef_.copy()
assert_array_almost_equal(coef_auto, coef_manual)
def test_nystroem_approximation():
# some basic tests
rnd = np.random.RandomState(0)
X = rnd.uniform(size=(10, 4))
# With n_components = n_samples this is exact
X_transformed = Nystroem(n_components=X.shape[0]).fit_transform(X)
K = rbf_kernel(X)
assert_array_almost_equal(np.dot(X_transformed, X_transformed.T), K)
trans = Nystroem(n_components=2, random_state=rnd)
X_transformed = trans.fit(X).transform(X)
assert_equal(X_transformed.shape, (X.shape[0], 2))
# test callable kernel
linear_kernel = lambda X, Y: np.dot(X, Y.T)
trans = Nystroem(n_components=2, kernel=linear_kernel, random_state=rnd)
X_transformed = trans.fit(X).transform(X)
assert_equal(X_transformed.shape, (X.shape[0], 2))
# test that available kernels fit and transform
kernels_available = kernel_metrics()
for kern in kernels_available:
trans = Nystroem(n_components=2, kernel=kern, random_state=rnd)
X_transformed = trans.fit(X).transform(X)
assert_equal(X_transformed.shape, (X.shape[0], 2))
def check_pipeline_consistency(name, Estimator):
if name in ('CCA', 'LocallyLinearEmbedding', 'KernelPCA') and _is_32bit():
# Those transformers yield non-deterministic output when executed on
# a 32bit Python. The same transformers are stable on 64bit Python.
# FIXME: try to isolate a minimalistic reproduction case only depending
# scipy and/or maybe generate a test dataset that does not
# cause such unstable behaviors.
msg = name + ' is non deterministic on 32bit Python'
raise SkipTest(msg)
# check that make_pipeline(est) gives same score as est
X, y = make_blobs(n_samples=30, centers=[[0, 0, 0], [1, 1, 1]],
random_state=0, n_features=2, cluster_std=0.1)
X -= X.min()
y = multioutput_estimator_convert_y_2d(name, y)
estimator = Estimator()
set_fast_parameters(estimator)
set_random_state(estimator)
pipeline = make_pipeline(estimator)
estimator.fit(X, y)
pipeline.fit(X, y)
funcs = ["score", "fit_transform"]
for func_name in funcs:
func = getattr(estimator, func_name, None)
if func is not None:
func_pipeline = getattr(pipeline, func_name)
result = func(X, y)
result_pipe = func_pipeline(X, y)
assert_array_almost_equal(result, result_pipe)
def check_multioutput(name):
# Check estimators on multi-output problems.
X_train = [[-2, -1], [-1, -1], [-1, -2], [1, 1], [1, 2], [2, 1], [-2, 1],
[-1, 1], [-1, 2], [2, -1], [1, -1], [1, -2]]
y_train = [[-1, 0], [-1, 0], [-1, 0], [1, 1], [1, 1], [1, 1], [-1, 2],
[-1, 2], [-1, 2], [1, 3], [1, 3], [1, 3]]
X_test = [[-1, -1], [1, 1], [-1, 1], [1, -1]]
y_test = [[-1, 0], [1, 1], [-1, 2], [1, 3]]
est = FOREST_ESTIMATORS[name](random_state=0, bootstrap=False)
y_pred = est.fit(X_train, y_train).predict(X_test)
assert_array_almost_equal(y_pred, y_test)
if name in FOREST_CLASSIFIERS:
with np.errstate(divide="ignore"):
proba = est.predict_proba(X_test)
assert_equal(len(proba), 2)
assert_equal(proba[0].shape, (4, 2))
assert_equal(proba[1].shape, (4, 4))
log_proba = est.predict_log_proba(X_test)
assert_equal(len(log_proba), 2)
assert_equal(log_proba[0].shape, (4, 2))
assert_equal(log_proba[1].shape, (4, 4))
def test_binary_perplexity_stability():
# Binary perplexity search should be stable.
# The binary_search_perplexity had a bug wherein the P array
# was uninitialized, leading to sporadically failing tests.
k = 10
n_samples = 100
random_state = check_random_state(0)
distances = random_state.randn(n_samples, 2).astype(np.float32)
# Distances shouldn't be negative
distances = np.abs(distances.dot(distances.T))
np.fill_diagonal(distances, 0.0)
last_P = None
neighbors_nn = np.argsort(distances, axis=1)[:, :k].astype(np.int64)
for _ in range(100):
P = _binary_search_perplexity(distances.copy(), neighbors_nn.copy(),
3, verbose=0)
P1 = _joint_probabilities_nn(distances, neighbors_nn, 3, verbose=0)
# Convert the sparse matrix to a dense one for testing
P1 = P1.toarray()
if last_P is None:
last_P = P
last_P1 = P1
else:
assert_array_almost_equal(P, last_P, decimal=4)
assert_array_almost_equal(P1, last_P1, decimal=4)
def test_multinomial_grad_hess():
rng = np.random.RandomState(0)
n_samples, n_features, n_classes = 100, 5, 3
X = rng.randn(n_samples, n_features)
w = rng.rand(n_classes, n_features)
Y = np.zeros((n_samples, n_classes))
ind = np.argmax(np.dot(X, w.T), axis=1)
Y[range(0, n_samples), ind] = 1
w = w.ravel()
sample_weights = np.ones(X.shape[0])
grad, hessp = _multinomial_grad_hess(w, X, Y, alpha=1.,
sample_weight=sample_weights)
# extract first column of hessian matrix
vec = np.zeros(n_features * n_classes)
vec[0] = 1
hess_col = hessp(vec)
# Estimate hessian using least squares as done in
# test_logistic_grad_hess
e = 1e-3
d_x = np.linspace(-e, e, 30)
d_grad = np.array([
_multinomial_grad_hess(w + t * vec, X, Y, alpha=1.,
sample_weight=sample_weights)[0]
for t in d_x
])
d_grad -= d_grad.mean(axis=0)
approx_hess_col = linalg.lstsq(d_x[:, np.newaxis], d_grad)[0].ravel()
assert_array_almost_equal(hess_col, approx_hess_col)
def test_classification_sample_weight():
X = [[0], [0], [1]]
y = [0, 1, 0]
sample_weight = [0.1, 1., 0.1]
clf = DummyClassifier().fit(X, y, sample_weight)
assert_array_almost_equal(clf.class_prior_, [0.2 / 1.2, 1. / 1.2])
def test_intercept_logistic_helper():
n_samples, n_features = 10, 5
X, y = make_classification(n_samples=n_samples, n_features=n_features,
random_state=0)
# Fit intercept case.
alpha = 1.
w = np.ones(n_features + 1)
grad_interp, hess_interp = _logistic_grad_hess(w, X, y, alpha)
loss_interp = _logistic_loss(w, X, y, alpha)
# Do not fit intercept. This can be considered equivalent to adding
# a feature vector of ones, i.e column of one vectors.
X_ = np.hstack((X, np.ones(10)[:, np.newaxis]))
grad, hess = _logistic_grad_hess(w, X_, y, alpha)
loss = _logistic_loss(w, X_, y, alpha)
# In the fit_intercept=False case, the feature vector of ones is
# penalized. This should be taken care of.
assert_almost_equal(loss_interp + 0.5 * (w[-1] ** 2), loss)
# Check gradient.
assert_array_almost_equal(grad_interp[:n_features], grad[:n_features])
assert_almost_equal(grad_interp[-1] + alpha * w[-1], grad[-1])
rng = np.random.RandomState(0)
grad = rng.rand(n_features + 1)
hess_interp = hess_interp(grad)
hess = hess(grad)
assert_array_almost_equal(hess_interp[:n_features], hess[:n_features])
assert_almost_equal(hess_interp[-1] + alpha * grad[-1], hess[-1])
def test_write_parameters():
# Test that we can write to coef_ and intercept_
clf = LogisticRegression(random_state=0)
clf.fit(X, Y1)
clf.coef_[:] = 0
clf.intercept_[:] = 0
assert_array_almost_equal(clf.decision_function(X), 0)
def check_regressors_int(name, Regressor):
X, _ = _boston_subset()
X = X[:50]
rnd = np.random.RandomState(0)
y = rnd.randint(3, size=X.shape[0])
y = multioutput_estimator_convert_y_2d(name, y)
rnd = np.random.RandomState(0)
# catch deprecation warnings
with warnings.catch_warnings(record=True):
# separate estimators to control random seeds
regressor_1 = Regressor()
regressor_2 = Regressor()
set_fast_parameters(regressor_1)
set_fast_parameters(regressor_2)
set_random_state(regressor_1)
set_random_state(regressor_2)
if name in CROSS_DECOMPOSITION:
y_ = np.vstack([y, 2 * y + rnd.randint(2, size=len(y))])
y_ = y_.T
else:
y_ = y
# fit
regressor_1.fit(X, y_)
pred1 = regressor_1.predict(X)
regressor_2.fit(X, y_.astype(np.float))
pred2 = regressor_2.predict(X)
assert_array_almost_equal(pred1, pred2, 2, name)
def test_logistic_regressioncv_class_weights():
X, y = make_classification(n_samples=20, n_features=20, n_informative=10,
n_classes=3, random_state=0)
# Test the liblinear fails when class_weight of type dict is
# provided, when it is multiclass. However it can handle
# binary problems.
clf_lib = LogisticRegressionCV(class_weight={0: 0.1, 1: 0.2},
solver='liblinear')
assert_raises(ValueError, clf_lib.fit, X, y)
y_ = y.copy()
y_[y == 2] = 1
clf_lib.fit(X, y_)
assert_array_equal(clf_lib.classes_, [0, 1])
# Test for class_weight=auto
X, y = make_classification(n_samples=20, n_features=20, n_informative=10,
random_state=0)
clf_lbf = LogisticRegressionCV(solver='lbfgs', fit_intercept=False,
class_weight='auto')
clf_lbf.fit(X, y)
clf_lib = LogisticRegressionCV(solver='liblinear', fit_intercept=False,
class_weight='auto')
clf_lib.fit(X, y)
assert_array_almost_equal(clf_lib.coef_, clf_lbf.coef_, decimal=4)
def test_predict_iris():
# Test logistic regression with the iris dataset
n_samples, n_features = iris.data.shape
target = iris.target_names[iris.target]
# Test that both multinomial and OvR solvers handle
# multiclass data correctly and give good accuracy
# score (>0.95) for the training data.
for clf in [LogisticRegression(C=len(iris.data)),
LogisticRegression(C=len(iris.data), solver='lbfgs',
multi_class='multinomial'),
LogisticRegression(C=len(iris.data), solver='newton-cg',
multi_class='multinomial'),
LogisticRegression(C=len(iris.data), solver='sag', tol=1e-2,
multi_class='ovr', random_state=42),
LogisticRegression(C=len(iris.data), solver='saga', tol=1e-2,
multi_class='ovr', random_state=42)
]:
clf.fit(iris.data, target)
assert_array_equal(np.unique(target), clf.classes_)
pred = clf.predict(iris.data)
assert_greater(np.mean(pred == target), .95)
probabilities = clf.predict_proba(iris.data)
assert_array_almost_equal(probabilities.sum(axis=1),
np.ones(n_samples))
pred = iris.target_names[probabilities.argmax(axis=1)]
assert_greater(np.mean(pred == target), .95)
def test_consistency_path():
"""Test that the path algorithm is consistent"""
rng = np.random.RandomState(0)
X = np.concatenate((rng.randn(100, 2) + [1, 1], rng.randn(100, 2)))
y = [1] * 100 + [-1] * 100
Cs = np.logspace(0, 4, 10)
f = ignore_warnings
# can't test with fit_intercept=True since LIBLINEAR
# penalizes the intercept
for method in ('lbfgs', 'newton-cg', 'liblinear'):
coefs, Cs = f(logistic_regression_path)(
X, y, Cs=Cs, fit_intercept=False, tol=1e-16, solver=method)
for i, C in enumerate(Cs):
lr = LogisticRegression(C=C, fit_intercept=False, tol=1e-16)
lr.fit(X, y)
lr_coef = lr.coef_.ravel()
assert_array_almost_equal(lr_coef, coefs[i], decimal=4)
# test for fit_intercept=True
for method in ('lbfgs', 'newton-cg', 'liblinear'):
Cs = [1e3]
coefs, Cs = f(logistic_regression_path)(
X, y, Cs=Cs, fit_intercept=True, tol=1e-4, solver=method)
lr = LogisticRegression(C=Cs[0], fit_intercept=True, tol=1e-4,
intercept_scaling=10000)
lr.fit(X, y)
lr_coef = np.concatenate([lr.coef_.ravel(), lr.intercept_])
assert_array_almost_equal(lr_coef, coefs[0], decimal=4)
def test_liblinear_random_state():
X, y = make_classification(n_samples=20)
lr1 = LogisticRegression(random_state=0)
lr1.fit(X, y)
lr2 = LogisticRegression(random_state=0)
lr2.fit(X, y)
assert_array_almost_equal(lr1.coef_, lr2.coef_)
def test_logistic_regression_class_weights():
# Multinomial case: remove 90% of class 0
X = iris.data[45:, :]
y = iris.target[45:]
solvers = ("lbfgs", "newton-cg")
class_weight_dict = _compute_class_weight_dictionary(y)
for solver in solvers:
clf1 = LogisticRegression(solver=solver, multi_class="multinomial",
class_weight="balanced")
clf2 = LogisticRegression(solver=solver, multi_class="multinomial",
class_weight=class_weight_dict)
clf1.fit(X, y)
clf2.fit(X, y)
assert_array_almost_equal(clf1.coef_, clf2.coef_, decimal=4)
# Binary case: remove 90% of class 0 and 100% of class 2
X = iris.data[45:100, :]
y = iris.target[45:100]
solvers = ("lbfgs", "newton-cg", "liblinear")
class_weight_dict = _compute_class_weight_dictionary(y)
for solver in solvers:
clf1 = LogisticRegression(solver=solver, multi_class="ovr",
class_weight="balanced")
clf2 = LogisticRegression(solver=solver, multi_class="ovr",
class_weight=class_weight_dict)
clf1.fit(X, y)
clf2.fit(X, y)
assert_array_almost_equal(clf1.coef_, clf2.coef_, decimal=6)
def _check_transformer(name, Transformer, X, y):
if name in ('CCA', 'LocallyLinearEmbedding', 'KernelPCA') and _is_32bit():
# Those transformers yield non-deterministic output when executed on
# a 32bit Python. The same transformers are stable on 64bit Python.
# FIXME: try to isolate a minimalistic reproduction case only depending
# on numpy & scipy and/or maybe generate a test dataset that does not
# cause such unstable behaviors.
msg = name + ' is non deterministic on 32bit Python'
raise SkipTest(msg)
n_samples, n_features = np.asarray(X).shape
# catch deprecation warnings
with warnings.catch_warnings(record=True):
transformer = Transformer()
set_random_state(transformer)
if name == "KernelPCA":
transformer.remove_zero_eig = False
set_fast_parameters(transformer)
# fit
if name in CROSS_DECOMPOSITION:
y_ = np.c_[y, y]
y_[::2, 1] *= 2
else:
y_ = y
transformer.fit(X, y_)
X_pred = transformer.fit_transform(X, y=y_)
if isinstance(X_pred, tuple):
for x_pred in X_pred:
assert_equal(x_pred.shape[0], n_samples)
else:
assert_equal(X_pred.shape[0], n_samples)
if hasattr(transformer, 'transform'):
if name in CROSS_DECOMPOSITION:
X_pred2 = transformer.transform(X, y_)
X_pred3 = transformer.fit_transform(X, y=y_)
else:
X_pred2 = transformer.transform(X)
X_pred3 = transformer.fit_transform(X, y=y_)
if isinstance(X_pred, tuple) and isinstance(X_pred2, tuple):
for x_pred, x_pred2, x_pred3 in zip(X_pred, X_pred2, X_pred3):
assert_array_almost_equal(
x_pred, x_pred2, 2,
"fit_transform and transform outcomes not consistent in %s"
% Transformer)
assert_array_almost_equal(
x_pred, x_pred3, 2,
"consecutive fit_transform outcomes not consistent in %s" %
Transformer)
else:
assert_array_almost_equal(
X_pred, X_pred2, 2,
"fit_transform and transform outcomes not consistent in %s" %
Transformer)
assert_array_almost_equal(
X_pred, X_pred3, 2,
"consecutive fit_transform outcomes not consistent in %s" %
Transformer)
# raises error on malformed input for transform
if hasattr(X, 'T'):
# If it's not an array, it does not have a 'T' property
assert_raises(ValueError, transformer.transform, X.T)
def test_linear_kernel():
rng = np.random.RandomState(0)
X = rng.random_sample((5, 4))
K = linear_kernel(X, X)
# the diagonal elements of a linear kernel are their squared norm
assert_array_almost_equal(K.flat[::6], [linalg.norm(x)**2 for x in X])
def test_roc_curve_toydata():
# Binary classification
y_true = [0, 1]
y_score = [0, 1]
tpr, fpr, _ = roc_curve(y_true, y_score)
roc_auc = roc_auc_score(y_true, y_score)
assert_array_almost_equal(tpr, [0, 0, 1])
assert_array_almost_equal(fpr, [0, 1, 1])
assert_almost_equal(roc_auc, 1.)
y_true = [0, 1]
y_score = [1, 0]
tpr, fpr, _ = roc_curve(y_true, y_score)
roc_auc = roc_auc_score(y_true, y_score)
assert_array_almost_equal(tpr, [0, 1, 1])
assert_array_almost_equal(fpr, [0, 0, 1])
assert_almost_equal(roc_auc, 0.)
y_true = [1, 0]
y_score = [1, 1]
tpr, fpr, _ = roc_curve(y_true, y_score)
roc_auc = roc_auc_score(y_true, y_score)
assert_array_almost_equal(tpr, [0, 1])
assert_array_almost_equal(fpr, [0, 1])
assert_almost_equal(roc_auc, 0.5)
y_true = [1, 0]
y_score = [1, 0]
tpr, fpr, _ = roc_curve(y_true, y_score)
roc_auc = roc_auc_score(y_true, y_score)
assert_array_almost_equal(tpr, [0, 0, 1])
assert_array_almost_equal(fpr, [0, 1, 1])
assert_almost_equal(roc_auc, 1.)
y_true = [1, 0]
y_score = [0.5, 0.5]
tpr, fpr, _ = roc_curve(y_true, y_score)
roc_auc = roc_auc_score(y_true, y_score)
assert_array_almost_equal(tpr, [0, 1])
assert_array_almost_equal(fpr, [0, 1])
assert_almost_equal(roc_auc, .5)
y_true = [0, 0]
y_score = [0.25, 0.75]
# assert UndefinedMetricWarning because of no positive sample in y_true
tpr, fpr, _ = assert_warns(UndefinedMetricWarning, roc_curve, y_true,
y_score)
assert_raises(ValueError, roc_auc_score, y_true, y_score)
assert_array_almost_equal(tpr, [0., 0.5, 1.])
assert_array_almost_equal(fpr, [np.nan, np.nan, np.nan])
y_true = [1, 1]
y_score = [0.25, 0.75]
# assert UndefinedMetricWarning because of no negative sample in y_true
tpr, fpr, _ = assert_warns(UndefinedMetricWarning, roc_curve, y_true,
y_score)
assert_raises(ValueError, roc_auc_score, y_true, y_score)
assert_array_almost_equal(tpr, [np.nan, np.nan, np.nan])
assert_array_almost_equal(fpr, [0., 0.5, 1.])
# Multi-label classification task
y_true = np.array([[0, 1], [0, 1]])
y_score = np.array([[0, 1], [0, 1]])
assert_raises(ValueError, roc_auc_score, y_true, y_score, average="macro")
assert_raises(ValueError,
roc_auc_score,
y_true,
y_score,
average="weighted")
assert_almost_equal(roc_auc_score(y_true, y_score, average="samples"), 1.)
assert_almost_equal(roc_auc_score(y_true, y_score, average="micro"), 1.)
y_true = np.array([[0, 1], [0, 1]])
y_score = np.array([[0, 1], [1, 0]])
assert_raises(ValueError, roc_auc_score, y_true, y_score, average="macro")
assert_raises(ValueError,
roc_auc_score,
y_true,
y_score,
average="weighted")
assert_almost_equal(roc_auc_score(y_true, y_score, average="samples"), 0.5)
assert_almost_equal(roc_auc_score(y_true, y_score, average="micro"), 0.5)
y_true = np.array([[1, 0], [0, 1]])
y_score = np.array([[0, 1], [1, 0]])
assert_almost_equal(roc_auc_score(y_true, y_score, average="macro"), 0)
assert_almost_equal(roc_auc_score(y_true, y_score, average="weighted"), 0)
assert_almost_equal(roc_auc_score(y_true, y_score, average="samples"), 0)
assert_almost_equal(roc_auc_score(y_true, y_score, average="micro"), 0)
y_true = np.array([[1, 0], [0, 1]])
y_score = np.array([[0.5, 0.5], [0.5, 0.5]])
assert_almost_equal(roc_auc_score(y_true, y_score, average="macro"), .5)
assert_almost_equal(roc_auc_score(y_true, y_score, average="weighted"), .5)
assert_almost_equal(roc_auc_score(y_true, y_score, average="samples"), .5)
assert_almost_equal(roc_auc_score(y_true, y_score, average="micro"), .5)
def test_ward_linkage_tree_return_distance():
# Test return_distance option on linkage and ward trees
# test that return_distance when set true, gives same
# output on both structured and unstructured clustering.
n, p = 10, 5
rng = np.random.RandomState(0)
connectivity = np.ones((n, n))
for i in range(5):
X = .1 * rng.normal(size=(n, p))
X -= 4. * np.arange(n)[:, np.newaxis]
X -= X.mean(axis=1)[:, np.newaxis]
out_unstructured = ward_tree(X, return_distance=True)
out_structured = ward_tree(X,
connectivity=connectivity,
return_distance=True)
# get children
children_unstructured = out_unstructured[0]
children_structured = out_structured[0]
# check if we got the same clusters
assert_array_equal(children_unstructured, children_structured)
# check if the distances are the same
dist_unstructured = out_unstructured[-1]
dist_structured = out_structured[-1]
assert_array_almost_equal(dist_unstructured, dist_structured)
for linkage in ['average', 'complete']:
structured_items = linkage_tree(X,
connectivity=connectivity,
linkage=linkage,
return_distance=True)[-1]
unstructured_items = linkage_tree(X,
linkage=linkage,
return_distance=True)[-1]
structured_dist = structured_items[-1]
unstructured_dist = unstructured_items[-1]
structured_children = structured_items[0]
unstructured_children = unstructured_items[0]
assert_array_almost_equal(structured_dist, unstructured_dist)
assert_array_almost_equal(structured_children,
unstructured_children)
# test on the following dataset where we know the truth
# taken from scipy/cluster/tests/hierarchy_test_data.py
X = np.array([[1.43054825, -7.5693489], [6.95887839, 6.82293382],
[2.87137846, -9.68248579], [7.87974764, -6.05485803],
[8.24018364, -6.09495602], [7.39020262, 8.54004355]])
# truth
linkage_X_ward = np.array([[3., 4., 0.36265956, 2.],
[1., 5., 1.77045373, 2.],
[0., 2., 2.55760419, 2.],
[6., 8., 9.10208346, 4.],
[7., 9., 24.7784379, 6.]])
linkage_X_complete = np.array([[3., 4., 0.36265956, 2.],
[1., 5., 1.77045373, 2.],
[0., 2., 2.55760419, 2.],
[6., 8., 6.96742194, 4.],
[7., 9., 18.77445997, 6.]])
linkage_X_average = np.array([[3., 4., 0.36265956, 2.],
[1., 5., 1.77045373, 2.],
[0., 2., 2.55760419, 2.],
[6., 8., 6.55832839, 4.],
[7., 9., 15.44089605, 6.]])
n_samples, n_features = np.shape(X)
connectivity_X = np.ones((n_samples, n_samples))
out_X_unstructured = ward_tree(X, return_distance=True)
out_X_structured = ward_tree(X,
connectivity=connectivity_X,
return_distance=True)
# check that the labels are the same
assert_array_equal(linkage_X_ward[:, :2], out_X_unstructured[0])
assert_array_equal(linkage_X_ward[:, :2], out_X_structured[0])
# check that the distances are correct
assert_array_almost_equal(linkage_X_ward[:, 2], out_X_unstructured[4])
assert_array_almost_equal(linkage_X_ward[:, 2], out_X_structured[4])
linkage_options = ['complete', 'average']
X_linkage_truth = [linkage_X_complete, linkage_X_average]
for (linkage, X_truth) in zip(linkage_options, X_linkage_truth):
out_X_unstructured = linkage_tree(X,
return_distance=True,
linkage=linkage)
out_X_structured = linkage_tree(X,
connectivity=connectivity_X,
linkage=linkage,
return_distance=True)
# check that the labels are the same
assert_array_equal(X_truth[:, :2], out_X_unstructured[0])
assert_array_equal(X_truth[:, :2], out_X_structured[0])
# check that the distances are correct
assert_array_almost_equal(X_truth[:, 2], out_X_unstructured[4])
assert_array_almost_equal(X_truth[:, 2], out_X_structured[4])
def test_multitarget():
"""
Assure that estimators receiving multidimensional y do the right thing
"""
X = diabetes.data
Y = np.vstack([diabetes.target, diabetes.target ** 2]).T
n_targets = Y.shape[1]
for estimator in (linear_model.LassoLars(), linear_model.Lars()):
estimator.fit(X, Y)
Y_pred = estimator.predict(X)
Y_dec = estimator.decision_function(X)
assert_array_almost_equal(Y_pred, Y_dec)
alphas, active, coef, path = (estimator.alphas_, estimator.active_,
estimator.coef_, estimator.coef_path_)
for k in range(n_targets):
estimator.fit(X, Y[:, k])
y_pred = estimator.predict(X)
assert_array_almost_equal(alphas[k], estimator.alphas_)
assert_array_almost_equal(active[k], estimator.active_)
assert_array_almost_equal(coef[k], estimator.coef_)
assert_array_almost_equal(path[k], estimator.coef_path_)
assert_array_almost_equal(Y_pred[:, k], y_pred)
def test_pls():
d = load_linnerud()
X = d.data
Y = d.target
# 1) Canonical (symmetric) PLS (PLS 2 blocks canonical mode A)
# ===========================================================
# Compare 2 algo.: nipals vs. svd
# ------------------------------
pls_bynipals = pls_.PLSCanonical(n_components=X.shape[1])
pls_bynipals.fit(X, Y)
pls_bysvd = pls_.PLSCanonical(algorithm="svd", n_components=X.shape[1])
pls_bysvd.fit(X, Y)
# check equalities of loading (up to the sign of the second column)
assert_array_almost_equal(
pls_bynipals.x_loadings_,
np.multiply(pls_bysvd.x_loadings_, np.array([1, -1, 1])),
decimal=5,
err_msg="nipals and svd implementation lead to different x loadings")
assert_array_almost_equal(
pls_bynipals.y_loadings_,
np.multiply(pls_bysvd.y_loadings_, np.array([1, -1, 1])),
decimal=5,
err_msg="nipals and svd implementation lead to different y loadings")
# Check PLS properties (with n_components=X.shape[1])
# ---------------------------------------------------
plsca = pls_.PLSCanonical(n_components=X.shape[1])
plsca.fit(X, Y)
T = plsca.x_scores_
P = plsca.x_loadings_
Wx = plsca.x_weights_
U = plsca.y_scores_
Q = plsca.y_loadings_
Wy = plsca.y_weights_
def check_ortho(M, err_msg):
K = np.dot(M.T, M)
assert_array_almost_equal(K, np.diag(np.diag(K)), err_msg=err_msg)
# Orthogonality of weights
# ~~~~~~~~~~~~~~~~~~~~~~~~
check_ortho(Wx, "x weights are not orthogonal")
check_ortho(Wy, "y weights are not orthogonal")
# Orthogonality of latent scores
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
check_ortho(T, "x scores are not orthogonal")
check_ortho(U, "y scores are not orthogonal")
# Check X = TP' and Y = UQ' (with (p == q) components)
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
# center scale X, Y
Xc, Yc, x_mean, y_mean, x_std, y_std =\
pls_._center_scale_xy(X.copy(), Y.copy(), scale=True)
assert_array_almost_equal(Xc, np.dot(T, P.T), err_msg="X != TP'")
assert_array_almost_equal(Yc, np.dot(U, Q.T), err_msg="Y != UQ'")
# Check that rotations on training data lead to scores
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Xr = plsca.transform(X)
assert_array_almost_equal(Xr,
plsca.x_scores_,
err_msg="rotation on X failed")
Xr, Yr = plsca.transform(X, Y)
assert_array_almost_equal(Xr,
plsca.x_scores_,
err_msg="rotation on X failed")
assert_array_almost_equal(Yr,
plsca.y_scores_,
err_msg="rotation on Y failed")
# "Non regression test" on canonical PLS
# --------------------------------------
# The results were checked against the R-package plspm
pls_ca = pls_.PLSCanonical(n_components=X.shape[1])
pls_ca.fit(X, Y)
x_weights = np.array([[-0.61330704, 0.25616119, -0.74715187],
[-0.74697144, 0.11930791, 0.65406368],
[-0.25668686, -0.95924297, -0.11817271]])
assert_array_almost_equal(pls_ca.x_weights_, x_weights)
x_rotations = np.array([[-0.61330704, 0.41591889, -0.62297525],
[-0.74697144, 0.31388326, 0.77368233],
[-0.25668686, -0.89237972, -0.24121788]])
assert_array_almost_equal(pls_ca.x_rotations_, x_rotations)
y_weights = np.array([[+0.58989127, 0.7890047, 0.1717553],
[+0.77134053, -0.61351791, 0.16920272],
[-0.23887670, -0.03267062, 0.97050016]])
assert_array_almost_equal(pls_ca.y_weights_, y_weights)
y_rotations = np.array([[+0.58989127, 0.7168115, 0.30665872],
[+0.77134053, -0.70791757, 0.19786539],
[-0.23887670, -0.00343595, 0.94162826]])
assert_array_almost_equal(pls_ca.y_rotations_, y_rotations)
# 2) Regression PLS (PLS2): "Non regression test"
# ===============================================
# The results were checked against the R-packages plspm, misOmics and pls
pls_2 = pls_.PLSRegression(n_components=X.shape[1])
pls_2.fit(X, Y)
x_weights = np.array([[-0.61330704, -0.00443647, 0.78983213],
[-0.74697144, -0.32172099, -0.58183269],
[-0.25668686, 0.94682413, -0.19399983]])
assert_array_almost_equal(pls_2.x_weights_, x_weights)
x_loadings = np.array([[-0.61470416, -0.24574278, 0.78983213],
[-0.65625755, -0.14396183, -0.58183269],
[-0.51733059, 1.00609417, -0.19399983]])
assert_array_almost_equal(pls_2.x_loadings_, x_loadings)
y_weights = np.array([[+0.32456184, 0.29892183, 0.20316322],
[+0.42439636, 0.61970543, 0.19320542],
[-0.13143144, -0.26348971, -0.17092916]])
assert_array_almost_equal(pls_2.y_weights_, y_weights)
y_loadings = np.array([[+0.32456184, 0.29892183, 0.20316322],
[+0.42439636, 0.61970543, 0.19320542],
[-0.13143144, -0.26348971, -0.17092916]])
assert_array_almost_equal(pls_2.y_loadings_, y_loadings)
# 3) Another non-regression test of Canonical PLS on random dataset
# =================================================================
# The results were checked against the R-package plspm
n = 500
p_noise = 10
q_noise = 5
# 2 latents vars:
np.random.seed(11)
l1 = np.random.normal(size=n)
l2 = np.random.normal(size=n)
latents = np.array([l1, l1, l2, l2]).T
X = latents + np.random.normal(size=4 * n).reshape((n, 4))
Y = latents + np.random.normal(size=4 * n).reshape((n, 4))
X = np.concatenate(
(X, np.random.normal(size=p_noise * n).reshape(n, p_noise)), axis=1)
Y = np.concatenate(
(Y, np.random.normal(size=q_noise * n).reshape(n, q_noise)), axis=1)
np.random.seed(None)
pls_ca = pls_.PLSCanonical(n_components=3)
pls_ca.fit(X, Y)
x_weights = np.array([[0.65803719, 0.19197924, 0.21769083],
[0.7009113, 0.13303969, -0.15376699],
[0.13528197, -0.68636408, 0.13856546],
[0.16854574, -0.66788088, -0.12485304],
[-0.03232333, -0.04189855, 0.40690153],
[0.1148816, -0.09643158, 0.1613305],
[0.04792138, -0.02384992, 0.17175319],
[-0.06781, -0.01666137, -0.18556747],
[-0.00266945, -0.00160224, 0.11893098],
[-0.00849528, -0.07706095, 0.1570547],
[-0.00949471, -0.02964127, 0.34657036],
[-0.03572177, 0.0945091, 0.3414855],
[0.05584937, -0.02028961, -0.57682568],
[0.05744254, -0.01482333, -0.17431274]])
assert_array_almost_equal(pls_ca.x_weights_, x_weights)
x_loadings = np.array([[0.65649254, 0.1847647, 0.15270699],
[0.67554234, 0.15237508, -0.09182247],
[0.19219925, -0.67750975, 0.08673128],
[0.2133631, -0.67034809, -0.08835483],
[-0.03178912, -0.06668336, 0.43395268],
[0.15684588, -0.13350241, 0.20578984],
[0.03337736, -0.03807306, 0.09871553],
[-0.06199844, 0.01559854, -0.1881785],
[0.00406146, -0.00587025, 0.16413253],
[-0.00374239, -0.05848466, 0.19140336],
[0.00139214, -0.01033161, 0.32239136],
[-0.05292828, 0.0953533, 0.31916881],
[0.04031924, -0.01961045, -0.65174036],
[0.06172484, -0.06597366, -0.1244497]])
assert_array_almost_equal(pls_ca.x_loadings_, x_loadings)
y_weights = np.array([[0.66101097, 0.18672553, 0.22826092],
[0.69347861, 0.18463471, -0.23995597],
[0.14462724, -0.66504085, 0.17082434],
[0.22247955, -0.6932605, -0.09832993],
[0.07035859, 0.00714283, 0.67810124],
[0.07765351, -0.0105204, -0.44108074],
[-0.00917056, 0.04322147, 0.10062478],
[-0.01909512, 0.06182718, 0.28830475],
[0.01756709, 0.04797666, 0.32225745]])
assert_array_almost_equal(pls_ca.y_weights_, y_weights)
y_loadings = np.array([[0.68568625, 0.1674376, 0.0969508],
[0.68782064, 0.20375837, -0.1164448],
[0.11712173, -0.68046903, 0.12001505],
[0.17860457, -0.6798319, -0.05089681],
[0.06265739, -0.0277703, 0.74729584],
[0.0914178, 0.00403751, -0.5135078],
[-0.02196918, -0.01377169, 0.09564505],
[-0.03288952, 0.09039729, 0.31858973],
[0.04287624, 0.05254676, 0.27836841]])
assert_array_almost_equal(pls_ca.y_loadings_, y_loadings)
# Orthogonality of weights
# ~~~~~~~~~~~~~~~~~~~~~~~~
check_ortho(pls_ca.x_weights_, "x weights are not orthogonal")
check_ortho(pls_ca.y_weights_, "y weights are not orthogonal")
# Orthogonality of latent scores
# ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
check_ortho(pls_ca.x_scores_, "x scores are not orthogonal")
check_ortho(pls_ca.y_scores_, "y scores are not orthogonal")
def check_ortho(M, err_msg):
K = np.dot(M.T, M)
assert_array_almost_equal(K, np.diag(np.diag(K)), err_msg=err_msg)
def test_rbf_kernel():
rng = np.random.RandomState(0)
X = rng.random_sample((5, 4))
K = rbf_kernel(X, X)
# the diagonal elements of a rbf kernel are 1
assert_array_almost_equal(K.flat[::6], np.ones(5))
def test_paired_manhattan_distances():
# Check the paired manhattan distances computation
X = [[0], [0]]
Y = [[1], [2]]
D = paired_manhattan_distances(X, Y)
assert_array_almost_equal(D, [1., 2.])
def test_kernel_symmetry(kernel):
# Valid kernels should be symmetric
rng = np.random.RandomState(0)
X = rng.random_sample((5, 4))
K = kernel(X, X)
assert_array_almost_equal(K, K.T, 15)
def test_pairwise_distances_argmin_min():
# Check pairwise minimum distances computation for any metric
X = [[0], [1]]
Y = [[-2], [3]]
Xsp = dok_matrix(X)
Ysp = csr_matrix(Y, dtype=np.float32)
expected_idx = [0, 1]
expected_vals = [2, 2]
expected_vals_sq = [4, 4]
# euclidean metric
idx, vals = pairwise_distances_argmin_min(X, Y, metric="euclidean")
idx2 = pairwise_distances_argmin(X, Y, metric="euclidean")
assert_array_almost_equal(idx, expected_idx)
assert_array_almost_equal(idx2, expected_idx)
assert_array_almost_equal(vals, expected_vals)
# sparse matrix case
idxsp, valssp = pairwise_distances_argmin_min(Xsp, Ysp, metric="euclidean")
assert_array_almost_equal(idxsp, expected_idx)
assert_array_almost_equal(valssp, expected_vals)
# We don't want np.matrix here
assert_equal(type(idxsp), np.ndarray)
assert_equal(type(valssp), np.ndarray)
# euclidean metric squared
idx, vals = pairwise_distances_argmin_min(X,
Y,
metric="euclidean",
metric_kwargs={"squared": True})
assert_array_almost_equal(idx, expected_idx)
assert_array_almost_equal(vals, expected_vals_sq)
# Non-euclidean scikit-learn metric
idx, vals = pairwise_distances_argmin_min(X, Y, metric="manhattan")
idx2 = pairwise_distances_argmin(X, Y, metric="manhattan")
assert_array_almost_equal(idx, expected_idx)
assert_array_almost_equal(idx2, expected_idx)
assert_array_almost_equal(vals, expected_vals)
# sparse matrix case
idxsp, valssp = pairwise_distances_argmin_min(Xsp, Ysp, metric="manhattan")
assert_array_almost_equal(idxsp, expected_idx)
assert_array_almost_equal(valssp, expected_vals)
# Non-euclidean Scipy distance (callable)
idx, vals = pairwise_distances_argmin_min(X,
Y,
metric=minkowski,
metric_kwargs={"p": 2})
assert_array_almost_equal(idx, expected_idx)
assert_array_almost_equal(vals, expected_vals)
# Non-euclidean Scipy distance (string)
idx, vals = pairwise_distances_argmin_min(X,
Y,
metric="minkowski",
metric_kwargs={"p": 2})
assert_array_almost_equal(idx, expected_idx)
assert_array_almost_equal(vals, expected_vals)
# Compare with naive implementation
rng = np.random.RandomState(0)
X = rng.randn(97, 149)
Y = rng.randn(111, 149)
dist = pairwise_distances(X, Y, metric="manhattan")
dist_orig_ind = dist.argmin(axis=0)
dist_orig_val = dist[dist_orig_ind, range(len(dist_orig_ind))]
dist_chunked_ind, dist_chunked_val = pairwise_distances_argmin_min(
X, Y, axis=0, metric="manhattan")
np.testing.assert_almost_equal(dist_orig_ind, dist_chunked_ind, decimal=7)
np.testing.assert_almost_equal(dist_orig_val, dist_chunked_val, decimal=7)
# Test batch_size deprecation warning
assert_warns_message(DeprecationWarning,
"version 0.22",
pairwise_distances_argmin_min,
X,
Y,
batch_size=500,
metric='euclidean')
def test_pairwise_distances():
# Test the pairwise_distance helper function.
rng = np.random.RandomState(0)
# Euclidean distance should be equivalent to calling the function.
X = rng.random_sample((5, 4))
S = pairwise_distances(X, metric="euclidean")
S2 = euclidean_distances(X)
assert_array_almost_equal(S, S2)
# Euclidean distance, with Y != X.
Y = rng.random_sample((2, 4))
S = pairwise_distances(X, Y, metric="euclidean")
S2 = euclidean_distances(X, Y)
assert_array_almost_equal(S, S2)
# Test with tuples as X and Y
X_tuples = tuple([tuple([v for v in row]) for row in X])
Y_tuples = tuple([tuple([v for v in row]) for row in Y])
S2 = pairwise_distances(X_tuples, Y_tuples, metric="euclidean")
assert_array_almost_equal(S, S2)
# "cityblock" uses scikit-learn metric, cityblock (function) is
# scipy.spatial.
S = pairwise_distances(X, metric="cityblock")
S2 = pairwise_distances(X, metric=cityblock)
assert_equal(S.shape[0], S.shape[1])
assert_equal(S.shape[0], X.shape[0])
assert_array_almost_equal(S, S2)
# The manhattan metric should be equivalent to cityblock.
S = pairwise_distances(X, Y, metric="manhattan")
S2 = pairwise_distances(X, Y, metric=cityblock)
assert_equal(S.shape[0], X.shape[0])
assert_equal(S.shape[1], Y.shape[0])
assert_array_almost_equal(S, S2)
# Test cosine as a string metric versus cosine callable
# The string "cosine" uses sklearn.metric,
# while the function cosine is scipy.spatial
S = pairwise_distances(X, Y, metric="cosine")
S2 = pairwise_distances(X, Y, metric=cosine)
assert_equal(S.shape[0], X.shape[0])
assert_equal(S.shape[1], Y.shape[0])
assert_array_almost_equal(S, S2)
# Test with sparse X and Y,
# currently only supported for Euclidean, L1 and cosine.
X_sparse = csr_matrix(X)
Y_sparse = csr_matrix(Y)
S = pairwise_distances(X_sparse, Y_sparse, metric="euclidean")
S2 = euclidean_distances(X_sparse, Y_sparse)
assert_array_almost_equal(S, S2)
S = pairwise_distances(X_sparse, Y_sparse, metric="cosine")
S2 = cosine_distances(X_sparse, Y_sparse)
assert_array_almost_equal(S, S2)
S = pairwise_distances(X_sparse, Y_sparse.tocsc(), metric="manhattan")
S2 = manhattan_distances(X_sparse.tobsr(), Y_sparse.tocoo())
assert_array_almost_equal(S, S2)
S2 = manhattan_distances(X, Y)
assert_array_almost_equal(S, S2)
# Test with scipy.spatial.distance metric, with a kwd
kwds = {"p": 2.0}
S = pairwise_distances(X, Y, metric="minkowski", **kwds)
S2 = pairwise_distances(X, Y, metric=minkowski, **kwds)
assert_array_almost_equal(S, S2)
# same with Y = None
kwds = {"p": 2.0}
S = pairwise_distances(X, metric="minkowski", **kwds)
S2 = pairwise_distances(X, metric=minkowski, **kwds)
assert_array_almost_equal(S, S2)
# Test that scipy distance metrics throw an error if sparse matrix given
assert_raises(TypeError, pairwise_distances, X_sparse, metric="minkowski")
assert_raises(TypeError,
pairwise_distances,
X,
Y_sparse,
metric="minkowski")
# Test that a value error is raised if the metric is unknown
assert_raises(ValueError, pairwise_distances, X, Y, metric="blah")
def test_precision_recall_curve_toydata():
with np.errstate(all="raise"):
# Binary classification
y_true = [0, 1]
y_score = [0, 1]
p, r, _ = precision_recall_curve(y_true, y_score)
auc_prc = average_precision_score(y_true, y_score)
assert_array_almost_equal(p, [1, 1])
assert_array_almost_equal(r, [1, 0])
assert_almost_equal(auc_prc, 1.)
y_true = [0, 1]
y_score = [1, 0]
p, r, _ = precision_recall_curve(y_true, y_score)
auc_prc = average_precision_score(y_true, y_score)
assert_array_almost_equal(p, [0.5, 0., 1.])
assert_array_almost_equal(r, [1., 0., 0.])
# Here we are doing a terrible prediction: we are always getting
# it wrong, hence the average_precision_score is the accuracy at
# chance: 50%
assert_almost_equal(auc_prc, 0.5)
y_true = [1, 0]
y_score = [1, 1]
p, r, _ = precision_recall_curve(y_true, y_score)
auc_prc = average_precision_score(y_true, y_score)
assert_array_almost_equal(p, [0.5, 1])
assert_array_almost_equal(r, [1., 0])
assert_almost_equal(auc_prc, .5)
y_true = [1, 0]
y_score = [1, 0]
p, r, _ = precision_recall_curve(y_true, y_score)
auc_prc = average_precision_score(y_true, y_score)
assert_array_almost_equal(p, [1, 1])
assert_array_almost_equal(r, [1, 0])
assert_almost_equal(auc_prc, 1.)
y_true = [1, 0]
y_score = [0.5, 0.5]
p, r, _ = precision_recall_curve(y_true, y_score)
auc_prc = average_precision_score(y_true, y_score)
assert_array_almost_equal(p, [0.5, 1])
assert_array_almost_equal(r, [1, 0.])
assert_almost_equal(auc_prc, .5)
y_true = [0, 0]
y_score = [0.25, 0.75]
assert_raises(Exception, precision_recall_curve, y_true, y_score)
assert_raises(Exception, average_precision_score, y_true, y_score)
y_true = [1, 1]
y_score = [0.25, 0.75]
p, r, _ = precision_recall_curve(y_true, y_score)
assert_almost_equal(average_precision_score(y_true, y_score), 1.)
assert_array_almost_equal(p, [1., 1., 1.])
assert_array_almost_equal(r, [1, 0.5, 0.])
# Multi-label classification task
y_true = np.array([[0, 1], [0, 1]])
y_score = np.array([[0, 1], [0, 1]])
assert_raises(Exception,
average_precision_score,
y_true,
y_score,
average="macro")
assert_raises(Exception,
average_precision_score,
y_true,
y_score,
average="weighted")
assert_almost_equal(
average_precision_score(y_true, y_score, average="samples"), 1.)
assert_almost_equal(
average_precision_score(y_true, y_score, average="micro"), 1.)
y_true = np.array([[0, 1], [0, 1]])
y_score = np.array([[0, 1], [1, 0]])
assert_raises(Exception,
average_precision_score,
y_true,
y_score,
average="macro")
assert_raises(Exception,
average_precision_score,
y_true,
y_score,
average="weighted")
assert_almost_equal(
average_precision_score(y_true, y_score, average="samples"), 0.75)
assert_almost_equal(
average_precision_score(y_true, y_score, average="micro"), 0.5)
y_true = np.array([[1, 0], [0, 1]])
y_score = np.array([[0, 1], [1, 0]])
assert_almost_equal(
average_precision_score(y_true, y_score, average="macro"), 0.5)
assert_almost_equal(
average_precision_score(y_true, y_score, average="weighted"), 0.5)
assert_almost_equal(
average_precision_score(y_true, y_score, average="samples"), 0.5)
assert_almost_equal(
average_precision_score(y_true, y_score, average="micro"), 0.5)
y_true = np.array([[1, 0], [0, 1]])
y_score = np.array([[0.5, 0.5], [0.5, 0.5]])
assert_almost_equal(
average_precision_score(y_true, y_score, average="macro"), 0.5)
assert_almost_equal(
average_precision_score(y_true, y_score, average="weighted"), 0.5)
assert_almost_equal(
average_precision_score(y_true, y_score, average="samples"), 0.5)
assert_almost_equal(
average_precision_score(y_true, y_score, average="micro"), 0.5)
def check_classifiers_train(name, Classifier):
X_m, y_m = make_blobs(random_state=0)
X_m, y_m = shuffle(X_m, y_m, random_state=7)
X_m = StandardScaler().fit_transform(X_m)
# generate binary problem from multi-class one
y_b = y_m[y_m != 2]
X_b = X_m[y_m != 2]
for (X, y) in [(X_m, y_m), (X_b, y_b)]:
# catch deprecation warnings
classes = np.unique(y)
n_classes = len(classes)
n_samples, n_features = X.shape
with warnings.catch_warnings(record=True):
classifier = Classifier()
if name in ['BernoulliNB', 'MultinomialNB']:
X -= X.min()
set_fast_parameters(classifier)
# raises error on malformed input for fit
assert_raises(ValueError, classifier.fit, X, y[:-1])
# fit
classifier.fit(X, y)
# with lists
classifier.fit(X.tolist(), y.tolist())
assert_true(hasattr(classifier, "classes_"))
y_pred = classifier.predict(X)
assert_equal(y_pred.shape, (n_samples, ))
# training set performance
if name not in ['BernoulliNB', 'MultinomialNB']:
assert_greater(accuracy_score(y, y_pred), 0.85)
# raises error on malformed input for predict
assert_raises(ValueError, classifier.predict, X.T)
if hasattr(classifier, "decision_function"):
try:
# decision_function agrees with predict:
decision = classifier.decision_function(X)
if n_classes is 2:
assert_equal(decision.shape, (n_samples, ))
dec_pred = (decision.ravel() > 0).astype(np.int)
assert_array_equal(dec_pred, y_pred)
if (n_classes is 3 and not isinstance(classifier, BaseLibSVM)):
# 1on1 of LibSVM works differently
assert_equal(decision.shape, (n_samples, n_classes))
assert_array_equal(np.argmax(decision, axis=1), y_pred)
# raises error on malformed input
assert_raises(ValueError, classifier.decision_function, X.T)
# raises error on malformed input for decision_function
assert_raises(ValueError, classifier.decision_function, X.T)
except NotImplementedError:
pass
if hasattr(classifier, "predict_proba"):
# predict_proba agrees with predict:
y_prob = classifier.predict_proba(X)
assert_equal(y_prob.shape, (n_samples, n_classes))
assert_array_equal(np.argmax(y_prob, axis=1), y_pred)
# check that probas for all classes sum to one
assert_array_almost_equal(np.sum(y_prob, axis=1),
np.ones(n_samples))
# raises error on malformed input
assert_raises(ValueError, classifier.predict_proba, X.T)
# raises error on malformed input for predict_proba
assert_raises(ValueError, classifier.predict_proba, X.T)
def test_mnnb():
"""Test Multinomial Naive Bayes classification.
This checks that MultinomialNB implements fit and predict and returns
correct values for a simple toy dataset.
"""
for X in [X2, scipy.sparse.csr_matrix(X2)]:
# Check the ability to predict the learning set.
clf = MultinomialNB()
assert_raises(ValueError, clf.fit, -X, y2)
y_pred = clf.fit(X, y2).predict(X)
assert_array_equal(y_pred, y2)
# Verify that np.log(clf.predict_proba(X)) gives the same results as
# clf.predict_log_proba(X)
y_pred_proba = clf.predict_proba(X)
y_pred_log_proba = clf.predict_log_proba(X)
assert_array_almost_equal(np.log(y_pred_proba), y_pred_log_proba, 8)
# Check that incremental fitting yields the same results
clf2 = MultinomialNB()
clf2.partial_fit(X[:2], y2[:2], classes=np.unique(y2))
clf2.partial_fit(X[2:5], y2[2:5])
clf2.partial_fit(X[5:], y2[5:])
y_pred2 = clf2.predict(X)
assert_array_equal(y_pred2, y2)
y_pred_proba2 = clf2.predict_proba(X)
y_pred_log_proba2 = clf2.predict_log_proba(X)
assert_array_almost_equal(np.log(y_pred_proba2), y_pred_log_proba2, 8)
assert_array_almost_equal(y_pred_proba2, y_pred_proba)
assert_array_almost_equal(y_pred_log_proba2, y_pred_log_proba)
# Partial fit on the whole data at once should be the same as fit too
clf3 = MultinomialNB()
clf3.partial_fit(X, y2, classes=np.unique(y2))
y_pred3 = clf3.predict(X)
assert_array_equal(y_pred3, y2)
y_pred_proba3 = clf3.predict_proba(X)
y_pred_log_proba3 = clf3.predict_log_proba(X)
assert_array_almost_equal(np.log(y_pred_proba3), y_pred_log_proba3, 8)
assert_array_almost_equal(y_pred_proba3, y_pred_proba)
assert_array_almost_equal(y_pred_log_proba3, y_pred_log_proba)
def test_paired_euclidean_distances():
""" Check the paired Euclidean distances computation"""
X = [[0], [0]]
Y = [[1], [2]]
D = paired_euclidean_distances(X, Y)
assert_array_almost_equal(D, [1., 2.])
def test_pairwise_kernels():
""" Test the pairwise_kernels helper function. """
def callable_rbf_kernel(x, y, **kwds):
""" Callable version of pairwise.rbf_kernel. """
K = rbf_kernel(np.atleast_2d(x), np.atleast_2d(y), **kwds)
return K
rng = np.random.RandomState(0)
X = rng.random_sample((5, 4))
Y = rng.random_sample((2, 4))
# Test with all metrics that should be in PAIRWISE_KERNEL_FUNCTIONS.
test_metrics = ["rbf", "sigmoid", "polynomial", "linear", "chi2",
"additive_chi2"]
for metric in test_metrics:
function = PAIRWISE_KERNEL_FUNCTIONS[metric]
# Test with Y=None
K1 = pairwise_kernels(X, metric=metric)
K2 = function(X)
assert_array_almost_equal(K1, K2)
# Test with Y=Y
K1 = pairwise_kernels(X, Y=Y, metric=metric)
K2 = function(X, Y=Y)
assert_array_almost_equal(K1, K2)
# Test with tuples as X and Y
X_tuples = tuple([tuple([v for v in row]) for row in X])
Y_tuples = tuple([tuple([v for v in row]) for row in Y])
K2 = pairwise_kernels(X_tuples, Y_tuples, metric=metric)
assert_array_almost_equal(K1, K2)
# Test with sparse X and Y
X_sparse = csr_matrix(X)
Y_sparse = csr_matrix(Y)
if metric in ["chi2", "additive_chi2"]:
# these don't support sparse matrices yet
assert_raises(ValueError, pairwise_kernels,
X_sparse, Y=Y_sparse, metric=metric)
continue
K1 = pairwise_kernels(X_sparse, Y=Y_sparse, metric=metric)
assert_array_almost_equal(K1, K2)
# Test with a callable function, with given keywords.
metric = callable_rbf_kernel
kwds = {}
kwds['gamma'] = 0.1
K1 = pairwise_kernels(X, Y=Y, metric=metric, **kwds)
K2 = rbf_kernel(X, Y=Y, **kwds)
assert_array_almost_equal(K1, K2)
# callable function, X=Y
K1 = pairwise_kernels(X, Y=X, metric=metric, **kwds)
K2 = rbf_kernel(X, Y=X, **kwds)
assert_array_almost_equal(K1, K2)
def test_pairwise_distances():
""" Test the pairwise_distance helper function. """
rng = np.random.RandomState(0)
# Euclidean distance should be equivalent to calling the function.
X = rng.random_sample((5, 4))
S = pairwise_distances(X, metric="euclidean")
S2 = euclidean_distances(X)
assert_array_almost_equal(S, S2)
# Euclidean distance, with Y != X.
Y = rng.random_sample((2, 4))
S = pairwise_distances(X, Y, metric="euclidean")
S2 = euclidean_distances(X, Y)
assert_array_almost_equal(S, S2)
# Test with tuples as X and Y
X_tuples = tuple([tuple([v for v in row]) for row in X])
Y_tuples = tuple([tuple([v for v in row]) for row in Y])
S2 = pairwise_distances(X_tuples, Y_tuples, metric="euclidean")
assert_array_almost_equal(S, S2)
# "cityblock" uses sklearn metric, cityblock (function) is scipy.spatial.
S = pairwise_distances(X, metric="cityblock")
S2 = pairwise_distances(X, metric=cityblock)
assert_equal(S.shape[0], S.shape[1])
assert_equal(S.shape[0], X.shape[0])
assert_array_almost_equal(S, S2)
# The manhattan metric should be equivalent to cityblock.
S = pairwise_distances(X, Y, metric="manhattan")
S2 = pairwise_distances(X, Y, metric=cityblock)
assert_equal(S.shape[0], X.shape[0])
assert_equal(S.shape[1], Y.shape[0])
assert_array_almost_equal(S, S2)
# manhattan does not support sparse matrices atm.
assert_raises(ValueError, pairwise_distances, csr_matrix(X),
metric="manhattan")
# Low-level function for manhattan can divide in blocks to avoid
# using too much memory during the broadcasting
S3 = manhattan_distances(X, Y, size_threshold=10)
assert_array_almost_equal(S, S3)
# Test cosine as a string metric versus cosine callable
# "cosine" uses sklearn metric, cosine (function) is scipy.spatial
S = pairwise_distances(X, Y, metric="cosine")
S2 = pairwise_distances(X, Y, metric=cosine)
assert_equal(S.shape[0], X.shape[0])
assert_equal(S.shape[1], Y.shape[0])
assert_array_almost_equal(S, S2)
# Tests that precomputed metric returns pointer to, and not copy of, X.
S = np.dot(X, X.T)
S2 = pairwise_distances(S, metric="precomputed")
assert_true(S is S2)
# Test with sparse X and Y,
# currently only supported for euclidean and cosine
X_sparse = csr_matrix(X)
Y_sparse = csr_matrix(Y)
S = pairwise_distances(X_sparse, Y_sparse, metric="euclidean")
S2 = euclidean_distances(X_sparse, Y_sparse)
assert_array_almost_equal(S, S2)
S = pairwise_distances(X_sparse, Y_sparse, metric="cosine")
S2 = cosine_distances(X_sparse, Y_sparse)
assert_array_almost_equal(S, S2)
# Test with scipy.spatial.distance metric, with a kwd
kwds = {"p": 2.0}
S = pairwise_distances(X, Y, metric="minkowski", **kwds)
S2 = pairwise_distances(X, Y, metric=minkowski, **kwds)
assert_array_almost_equal(S, S2)
# same with Y = None
kwds = {"p": 2.0}
S = pairwise_distances(X, metric="minkowski", **kwds)
S2 = pairwise_distances(X, metric=minkowski, **kwds)
assert_array_almost_equal(S, S2)
# Test that scipy distance metrics throw an error if sparse matrix given
assert_raises(TypeError, pairwise_distances, X_sparse, metric="minkowski")
assert_raises(TypeError, pairwise_distances, X, Y_sparse,
metric="minkowski")
# Test that a value error is raised if the metric is unkown
assert_raises(ValueError, pairwise_distances, X, Y, metric="blah")
def test_explained_variance():
# Check that PCA output has unit-variance
rng = np.random.RandomState(0)
n_samples = 100
n_features = 80
X = rng.randn(n_samples, n_features)
pca = PCA(n_components=2, svd_solver='full').fit(X)
apca = PCA(n_components=2, svd_solver='arpack', random_state=0).fit(X)
assert_array_almost_equal(pca.explained_variance_,
apca.explained_variance_, 1)
assert_array_almost_equal(pca.explained_variance_ratio_,
apca.explained_variance_ratio_, 3)
rpca = PCA(n_components=2, svd_solver='randomized', random_state=42).fit(X)
assert_array_almost_equal(pca.explained_variance_,
rpca.explained_variance_, 1)
assert_array_almost_equal(pca.explained_variance_ratio_,
rpca.explained_variance_ratio_, 1)
# compare to empirical variances
X_pca = pca.transform(X)
assert_array_almost_equal(pca.explained_variance_,
np.var(X_pca, axis=0))
X_pca = apca.transform(X)
assert_array_almost_equal(apca.explained_variance_,
np.var(X_pca, axis=0))
X_rpca = rpca.transform(X)
assert_array_almost_equal(rpca.explained_variance_, np.var(X_rpca, axis=0),
decimal=1)
# Same with correlated data
X = datasets.make_classification(n_samples, n_features,
n_informative=n_features-2,
random_state=rng)[0]
pca = PCA(n_components=2).fit(X)
rpca = PCA(n_components=2, svd_solver='randomized',
random_state=rng).fit(X)
assert_array_almost_equal(pca.explained_variance_ratio_,
rpca.explained_variance_ratio_, 5)
def test_pairwise_distances_argmin_min():
""" Check pairwise minimum distances computation for any metric"""
X = [[0], [1]]
Y = [[-1], [2]]
Xsp = dok_matrix(X)
Ysp = csr_matrix(Y)
# euclidean metric
D, E = pairwise_distances_argmin_min(X, Y, metric="euclidean")
D2 = pairwise_distances_argmin(X, Y, metric="euclidean")
assert_array_almost_equal(D, [0, 1])
assert_array_almost_equal(D2, [0, 1])
assert_array_almost_equal(D, [0, 1])
assert_array_almost_equal(E, [1., 1.])
# sparse matrix case
Dsp, Esp = pairwise_distances_argmin_min(Xsp, Ysp, metric="euclidean")
assert_array_equal(Dsp, D)
assert_array_equal(Esp, E)
# We don't want np.matrix here
assert_equal(type(Dsp), np.ndarray)
assert_equal(type(Esp), np.ndarray)
# Non-euclidean sklearn metric
D, E = pairwise_distances_argmin_min(X, Y, metric="manhattan")
D2 = pairwise_distances_argmin(X, Y, metric="manhattan")
assert_array_almost_equal(D, [0, 1])
assert_array_almost_equal(D2, [0, 1])
assert_array_almost_equal(E, [1., 1.])
# sparse matrix case
assert_raises(ValueError,
pairwise_distances_argmin_min, Xsp, Ysp, metric="manhattan")
# Non-euclidean Scipy distance (callable)
D, E = pairwise_distances_argmin_min(X, Y, metric=minkowski,
metric_kwargs={"p": 2})
assert_array_almost_equal(D, [0, 1])
assert_array_almost_equal(E, [1., 1.])
# Non-euclidean Scipy distance (string)
D, E = pairwise_distances_argmin_min(X, Y, metric="minkowski",
metric_kwargs={"p": 2})
assert_array_almost_equal(D, [0, 1])
assert_array_almost_equal(E, [1., 1.])
# Compare with naive implementation
rng = np.random.RandomState(0)
X = rng.randn(97, 149)
Y = rng.randn(111, 149)
dist = pairwise_distances(X, Y, metric="manhattan")
dist_orig_ind = dist.argmin(axis=0)
dist_orig_val = dist[dist_orig_ind, range(len(dist_orig_ind))]
dist_chunked_ind, dist_chunked_val = pairwise_distances_argmin_min(
X, Y, axis=0, metric="manhattan", batch_size=50)
np.testing.assert_almost_equal(dist_orig_ind, dist_chunked_ind, decimal=7)
np.testing.assert_almost_equal(dist_orig_val, dist_chunked_val, decimal=7)
def test_lasso_toy():
"""
Test Lasso on a toy example for various values of alpha.
When validating this against glmnet notice that glmnet divides it
against nobs.
"""
X = [[-1], [0], [1]]
Y = [-1, 0, 1] # just a straight line
T = [[2], [3], [4]] # test sample
clf = Lasso(alpha=1e-8)
clf.fit(X, Y)
pred = clf.predict(T)
assert_array_almost_equal(clf.coef_, [1])
assert_array_almost_equal(pred, [2, 3, 4])
assert_almost_equal(clf.dual_gap_, 0)
clf = Lasso(alpha=0.1)
clf.fit(X, Y)
pred = clf.predict(T)
assert_array_almost_equal(clf.coef_, [.85])
assert_array_almost_equal(pred, [1.7, 2.55, 3.4])
assert_almost_equal(clf.dual_gap_, 0)
clf = Lasso(alpha=0.5)
clf.fit(X, Y)
pred = clf.predict(T)
assert_array_almost_equal(clf.coef_, [.25])
assert_array_almost_equal(pred, [0.5, 0.75, 1.])
assert_almost_equal(clf.dual_gap_, 0)
clf = Lasso(alpha=1)
clf.fit(X, Y)
pred = clf.predict(T)
assert_array_almost_equal(clf.coef_, [.0])
assert_array_almost_equal(pred, [0, 0, 0])
assert_almost_equal(clf.dual_gap_, 0)
def test_singular_values():
# Check that the PCA output has the correct singular values
rng = np.random.RandomState(0)
n_samples = 100
n_features = 80
X = rng.randn(n_samples, n_features)
pca = PCA(n_components=2, svd_solver='full',
random_state=rng).fit(X)
apca = PCA(n_components=2, svd_solver='arpack',
random_state=rng).fit(X)
rpca = PCA(n_components=2, svd_solver='randomized',
random_state=rng).fit(X)
assert_array_almost_equal(pca.singular_values_, apca.singular_values_, 12)
assert_array_almost_equal(pca.singular_values_, rpca.singular_values_, 1)
assert_array_almost_equal(apca.singular_values_, rpca.singular_values_, 1)
# Compare to the Frobenius norm
X_pca = pca.transform(X)
X_apca = apca.transform(X)
X_rpca = rpca.transform(X)
assert_array_almost_equal(np.sum(pca.singular_values_**2.0),
np.linalg.norm(X_pca, "fro")**2.0, 12)
assert_array_almost_equal(np.sum(apca.singular_values_**2.0),
np.linalg.norm(X_apca, "fro")**2.0, 12)
assert_array_almost_equal(np.sum(rpca.singular_values_**2.0),
np.linalg.norm(X_rpca, "fro")**2.0, 0)
# Compare to the 2-norms of the score vectors
assert_array_almost_equal(pca.singular_values_,
np.sqrt(np.sum(X_pca**2.0, axis=0)), 12)
assert_array_almost_equal(apca.singular_values_,
np.sqrt(np.sum(X_apca**2.0, axis=0)), 12)
assert_array_almost_equal(rpca.singular_values_,
np.sqrt(np.sum(X_rpca**2.0, axis=0)), 2)
# Set the singular values and see what we get back
rng = np.random.RandomState(0)
n_samples = 100
n_features = 110
X = rng.randn(n_samples, n_features)
pca = PCA(n_components=3, svd_solver='full', random_state=rng)
apca = PCA(n_components=3, svd_solver='arpack', random_state=rng)
rpca = PCA(n_components=3, svd_solver='randomized', random_state=rng)
X_pca = pca.fit_transform(X)
X_pca /= np.sqrt(np.sum(X_pca**2.0, axis=0))
X_pca[:, 0] *= 3.142
X_pca[:, 1] *= 2.718
X_hat = np.dot(X_pca, pca.components_)
pca.fit(X_hat)
apca.fit(X_hat)
rpca.fit(X_hat)
assert_array_almost_equal(pca.singular_values_, [3.142, 2.718, 1.0], 14)
assert_array_almost_equal(apca.singular_values_, [3.142, 2.718, 1.0], 14)
assert_array_almost_equal(rpca.singular_values_, [3.142, 2.718, 1.0], 14)
def check_sparse_input(name, X, X_sparse, y):
ForestEstimator = FOREST_ESTIMATORS[name]
dense = ForestEstimator(random_state=0, max_depth=2).fit(X, y)
sparse = ForestEstimator(random_state=0, max_depth=2).fit(X_sparse, y)
assert_array_almost_equal(sparse.apply(X), dense.apply(X))
if name in FOREST_CLASSIFIERS or name in FOREST_REGRESSORS:
assert_array_almost_equal(sparse.predict(X), dense.predict(X))
assert_array_almost_equal(sparse.feature_importances_,
dense.feature_importances_)
if name in FOREST_CLASSIFIERS:
assert_array_almost_equal(sparse.predict_proba(X),
dense.predict_proba(X))
assert_array_almost_equal(sparse.predict_log_proba(X),
dense.predict_log_proba(X))
if name in FOREST_TRANSFORMERS:
assert_array_almost_equal(
sparse.transform(X).toarray(),
dense.transform(X).toarray())
assert_array_almost_equal(
sparse.fit_transform(X).toarray(),
dense.fit_transform(X).toarray())
def test_enet_toy():
"""
Test ElasticNet for various parameters of alpha and l1_ratio.
Actually, the parameters alpha = 0 should not be allowed. However,
we test it as a border case.
ElasticNet is tested with and without precomputed Gram matrix
"""
X = np.array([[-1.], [0.], [1.]])
Y = [-1, 0, 1] # just a straight line
T = [[2.], [3.], [4.]] # test sample
# this should be the same as lasso
clf = ElasticNet(alpha=1e-8, l1_ratio=1.0)
clf.fit(X, Y)
pred = clf.predict(T)
assert_array_almost_equal(clf.coef_, [1])
assert_array_almost_equal(pred, [2, 3, 4])
assert_almost_equal(clf.dual_gap_, 0)
clf = ElasticNet(alpha=0.5, l1_ratio=0.3, max_iter=100, precompute=False)
clf.fit(X, Y)
pred = clf.predict(T)
assert_array_almost_equal(clf.coef_, [0.50819], decimal=3)
assert_array_almost_equal(pred, [1.0163, 1.5245, 2.0327], decimal=3)
assert_almost_equal(clf.dual_gap_, 0)
clf.set_params(max_iter=100, precompute=True)
clf.fit(X, Y) # with Gram
pred = clf.predict(T)
assert_array_almost_equal(clf.coef_, [0.50819], decimal=3)
assert_array_almost_equal(pred, [1.0163, 1.5245, 2.0327], decimal=3)
assert_almost_equal(clf.dual_gap_, 0)
clf.set_params(max_iter=100, precompute=np.dot(X.T, X))
clf.fit(X, Y) # with Gram
pred = clf.predict(T)
assert_array_almost_equal(clf.coef_, [0.50819], decimal=3)
assert_array_almost_equal(pred, [1.0163, 1.5245, 2.0327], decimal=3)
assert_almost_equal(clf.dual_gap_, 0)
clf = ElasticNet(alpha=0.5, l1_ratio=0.5)
clf.fit(X, Y)
pred = clf.predict(T)
assert_array_almost_equal(clf.coef_, [0.45454], 3)
assert_array_almost_equal(pred, [0.9090, 1.3636, 1.8181], 3)
assert_almost_equal(clf.dual_gap_, 0)
def test_sgd_proba(self):
# Check SGD.predict_proba
# Hinge loss does not allow for conditional prob estimate.
# We cannot use the factory here, because it defines predict_proba
# anyway.
clf = SGDClassifier(loss="hinge", alpha=0.01, n_iter=10).fit(X, Y)
assert_false(hasattr(clf, "predict_proba"))
assert_false(hasattr(clf, "predict_log_proba"))
# log and modified_huber losses can output probability estimates
# binary case
for loss in ["log", "modified_huber"]:
clf = self.factory(loss="modified_huber", alpha=0.01, n_iter=10)
clf.fit(X, Y)
p = clf.predict_proba([[3, 2]])
assert_true(p[0, 1] > 0.5)
p = clf.predict_proba([[-1, -1]])
assert_true(p[0, 1] < 0.5)
p = clf.predict_log_proba([[3, 2]])
assert_true(p[0, 1] > p[0, 0])
p = clf.predict_log_proba([[-1, -1]])
assert_true(p[0, 1] < p[0, 0])
# log loss multiclass probability estimates
clf = self.factory(loss="log", alpha=0.01, n_iter=10).fit(X2, Y2)
d = clf.decision_function([[.1, -.1], [.3, .2]])
p = clf.predict_proba([[.1, -.1], [.3, .2]])
assert_array_equal(np.argmax(p, axis=1), np.argmax(d, axis=1))
assert_almost_equal(p[0].sum(), 1)
assert_true(np.all(p[0] >= 0))
p = clf.predict_proba([[-1, -1]])
d = clf.decision_function([[-1, -1]])
assert_array_equal(np.argsort(p[0]), np.argsort(d[0]))
l = clf.predict_log_proba([[3, 2]])
p = clf.predict_proba([[3, 2]])
assert_array_almost_equal(np.log(p), l)
l = clf.predict_log_proba([[-1, -1]])
p = clf.predict_proba([[-1, -1]])
assert_array_almost_equal(np.log(p), l)
# Modified Huber multiclass probability estimates; requires a separate
# test because the hard zero/one probabilities may destroy the
# ordering present in decision_function output.
clf = self.factory(loss="modified_huber", alpha=0.01, n_iter=10)
clf.fit(X2, Y2)
d = clf.decision_function([[3, 2]])
p = clf.predict_proba([[3, 2]])
if not isinstance(self, SparseSGDClassifierTestCase):
assert_equal(np.argmax(d, axis=1), np.argmax(p, axis=1))
else: # XXX the sparse test gets a different X2 (?)
assert_equal(np.argmin(d, axis=1), np.argmin(p, axis=1))
# the following sample produces decision_function values < -1,
# which would cause naive normalization to fail (see comment
# in SGDClassifier.predict_proba)
x = X.mean(axis=0)
d = clf.decision_function([x])
if np.all(d < -1): # XXX not true in sparse test case (why?)
p = clf.predict_proba([x])
assert_array_almost_equal(p[0], [1 / 3.] * 3)
def check_memory_layout(name, dtype):
# Check that it works no matter the memory layout
est = FOREST_ESTIMATORS[name](random_state=0, bootstrap=False)
# Nothing
X = np.asarray(iris.data, dtype=dtype)
y = iris.target
assert_array_almost_equal(est.fit(X, y).predict(X), y)
# C-order
X = np.asarray(iris.data, order="C", dtype=dtype)
y = iris.target
assert_array_almost_equal(est.fit(X, y).predict(X), y)
# F-order
X = np.asarray(iris.data, order="F", dtype=dtype)
y = iris.target
assert_array_almost_equal(est.fit(X, y).predict(X), y)
# Contiguous
X = np.ascontiguousarray(iris.data, dtype=dtype)
y = iris.target
assert_array_almost_equal(est.fit(X, y).predict(X), y)
if est.base_estimator.splitter in SPARSE_SPLITTERS:
# csr matrix
X = csr_matrix(iris.data, dtype=dtype)
y = iris.target
assert_array_almost_equal(est.fit(X, y).predict(X), y)
# csc_matrix
X = csc_matrix(iris.data, dtype=dtype)
y = iris.target
assert_array_almost_equal(est.fit(X, y).predict(X), y)
# coo_matrix
X = coo_matrix(iris.data, dtype=dtype)
y = iris.target
assert_array_almost_equal(est.fit(X, y).predict(X), y)
# Strided
X = np.asarray(iris.data[::3], dtype=dtype)
y = iris.target[::3]
assert_array_almost_equal(est.fit(X, y).predict(X), y)
def test_discrete_prior():
"""Test whether class priors are properly set. """
for cls in [BernoulliNB, MultinomialNB]:
clf = cls().fit(X2, y2)
assert_array_almost_equal(np.log(np.array([2, 2, 2]) / 6.0),
clf.class_log_prior_, 8)
def test_dump():
X_sparse, y_dense = load_svmlight_file(datafile)
X_dense = X_sparse.toarray()
y_sparse = sp.csr_matrix(y_dense)
# slicing a csr_matrix can unsort its .indices, so test that we sort
# those correctly
X_sliced = X_sparse[np.arange(X_sparse.shape[0])]
y_sliced = y_sparse[np.arange(y_sparse.shape[0])]
for X in (X_sparse, X_dense, X_sliced):
for y in (y_sparse, y_dense, y_sliced):
for zero_based in (True, False):
for dtype in [np.float32, np.float64, np.int32]:
f = BytesIO()
# we need to pass a comment to get the version info in;
# LibSVM doesn't grok comments so they're not put in by
# default anymore.
if sp.issparse(y) and y.shape[0] == 1:
# make sure y's shape is: (n_samples, n_labels)
# when it is sparse
y = y.T
dump_svmlight_file(X.astype(dtype),
y,
f,
comment="test",
zero_based=zero_based)
f.seek(0)
comment = f.readline()
try:
comment = str(comment, "utf-8")
except TypeError: # fails in Python 2.x
pass
assert_in("scikit-learn %s" % sklearn.__version__, comment)
comment = f.readline()
try:
comment = str(comment, "utf-8")
except TypeError: # fails in Python 2.x
pass
assert_in(["one", "zero"][zero_based] + "-based", comment)
X2, y2 = load_svmlight_file(f,
dtype=dtype,
zero_based=zero_based)
assert_equal(X2.dtype, dtype)
assert_array_equal(X2.sorted_indices().indices, X2.indices)
X2_dense = X2.toarray()
if dtype == np.float32:
# allow a rounding error at the last decimal place
assert_array_almost_equal(X_dense.astype(dtype),
X2_dense, 4)
assert_array_almost_equal(y_dense.astype(dtype), y2, 4)
else:
# allow a rounding error at the last decimal place
assert_array_almost_equal(X_dense.astype(dtype),
X2_dense, 15)
assert_array_almost_equal(y_dense.astype(dtype), y2,
15)
def test_lasso_lars_vs_R_implementation():
# Test that sklearn LassoLars implementation agrees with the LassoLars
# implementation available in R (lars library) under the following
# scenarios:
# 1) fit_intercept=False and normalize=False
# 2) fit_intercept=True and normalize=True
# Let's generate the data used in the bug report 7778
y = np.array(
[-6.45006793, -3.51251449, -8.52445396, 6.12277822, -19.42109366])
x = np.array(
[[0.47299829, 0, 0, 0, 0], [0.08239882, 0.85784863, 0, 0, 0],
[0.30114139, -0.07501577, 0.80895216, 0, 0],
[-0.01460346, -0.1015233, 0.0407278, 0.80338378, 0],
[-0.69363927, 0.06754067, 0.18064514, -0.0803561, 0.40427291]])
X = x.T
###########################################################################
# Scenario 1: Let's compare R vs sklearn when fit_intercept=False and
# normalize=False
###########################################################################
#
# The R result was obtained using the following code:
#
# library(lars)
# model_lasso_lars = lars(X, t(y), type="lasso", intercept=FALSE,
# trace=TRUE, normalize=FALSE)
# r = t(model_lasso_lars$beta)
#
r = np.array([[
0, 0, 0, 0, 0, -79.810362809499026, -83.528788732782829,
-83.777653739190711, -83.784156932888934, -84.033390591756657
], [0, 0, 0, 0, -0.476624256777266, 0, 0, 0, 0, 0.025219751009936],
[
0, -3.577397088285891, -4.702795355871871,
-7.016748621359461, -7.614898471899412,
-0.336938391359179, 0, 0, 0.001213370600853,
0.048162321585148
],
[
0, 0, 0, 2.231558436628169, 2.723267514525966,
2.811549786389614, 2.813766976061531, 2.817462468949557,
2.817368178703816, 2.816221090636795
],
[
0, 0, -1.218422599914637, -3.457726183014808,
-4.021304522060710, -45.827461592423745,
-47.776608869312305, -47.911561610746404,
-47.914845922736234, -48.039562334265717
]])
model_lasso_lars = linear_model.LassoLars(alpha=0,
fit_intercept=False,
normalize=False)
model_lasso_lars.fit(X, y)
skl_betas = model_lasso_lars.coef_path_
assert_array_almost_equal(r, skl_betas, decimal=12)
###########################################################################
###########################################################################
# Scenario 2: Let's compare R vs sklearn when fit_intercept=True and
# normalize=True
#
# Note: When normalize is equal to True, R returns the coefficients in
# their original units, that is, they are rescaled back, whereas sklearn
# does not do that, therefore, we need to do this step before comparing
# their results.
###########################################################################
#
# The R result was obtained using the following code:
#
# library(lars)
# model_lasso_lars2 = lars(X, t(y), type="lasso", intercept=TRUE,
# trace=TRUE, normalize=TRUE)
# r2 = t(model_lasso_lars2$beta)
r2 = np.array(
[[0, 0, 0, 0, 0], [0, 0, 0, 8.371887668009453, 19.463768371044026],
[0, 0, 0, 0, 9.901611055290553],
[
0, 7.495923132833733, 9.245133544334507, 17.389369207545062,
26.971656815643499
], [0, 0, -1.569380717440311, -5.924804108067312,
-7.996385265061972]])
model_lasso_lars2 = linear_model.LassoLars(alpha=0,
fit_intercept=True,
normalize=True)
model_lasso_lars2.fit(X, y)
skl_betas2 = model_lasso_lars2.coef_path_
# Let's rescale back the coefficients returned by sklearn before comparing
# against the R result (read the note above)
temp = X - np.mean(X, axis=0)
normx = np.sqrt(np.sum(temp**2, axis=0))
skl_betas2 /= normx[:, np.newaxis]
assert_array_almost_equal(r2, skl_betas2, decimal=12)
def test_fastica_simple(add_noise=False):
""" Test the FastICA algorithm on very simple data.
"""
rng = np.random.RandomState(0)
# scipy.stats uses the global RNG:
np.random.seed(0)
n_samples = 1000
# Generate two sources:
s1 = (2 * np.sin(np.linspace(0, 100, n_samples)) > 0) - 1
s2 = stats.t.rvs(1, size=n_samples)
s = np.c_[s1, s2].T
center_and_norm(s)
s1, s2 = s
# Mixing angle
phi = 0.6
mixing = np.array([[np.cos(phi), np.sin(phi)],
[np.sin(phi), -np.cos(phi)]])
m = np.dot(mixing, s)
if add_noise:
m += 0.1 * rng.randn(2, 1000)
center_and_norm(m)
# function as fun arg
def g_test(x):
return x ** 3, (3 * x ** 2).mean(axis=-1)
algos = ['parallel', 'deflation']
nls = ['logcosh', 'exp', 'cube', g_test]
whitening = [True, False]
for algo, nl, whiten in itertools.product(algos, nls, whitening):
if whiten:
k_, mixing_, s_ = fastica(m.T, fun=nl, algorithm=algo)
assert_raises(ValueError, fastica, m.T, fun=np.tanh,
algorithm=algo)
else:
X = PCA(n_components=2, whiten=True).fit_transform(m.T)
k_, mixing_, s_ = fastica(X, fun=nl, algorithm=algo, whiten=False)
assert_raises(ValueError, fastica, X, fun=np.tanh,
algorithm=algo)
s_ = s_.T
# Check that the mixing model described in the docstring holds:
if whiten:
assert_almost_equal(s_, np.dot(np.dot(mixing_, k_), m))
center_and_norm(s_)
s1_, s2_ = s_
# Check to see if the sources have been estimated
# in the wrong order
if abs(np.dot(s1_, s2)) > abs(np.dot(s1_, s1)):
s2_, s1_ = s_
s1_ *= np.sign(np.dot(s1_, s1))
s2_ *= np.sign(np.dot(s2_, s2))
# Check that we have estimated the original sources
if not add_noise:
assert_almost_equal(np.dot(s1_, s1) / n_samples, 1, decimal=2)
assert_almost_equal(np.dot(s2_, s2) / n_samples, 1, decimal=2)
else:
assert_almost_equal(np.dot(s1_, s1) / n_samples, 1, decimal=1)
assert_almost_equal(np.dot(s2_, s2) / n_samples, 1, decimal=1)
# Test FastICA class
_, _, sources_fun = fastica(m.T, fun=nl, algorithm=algo, random_state=0)
ica = FastICA(fun=nl, algorithm=algo, random_state=0)
sources = ica.fit_transform(m.T)
assert_equal(ica.components_.shape, (2, 2))
assert_equal(sources.shape, (1000, 2))
assert_array_almost_equal(sources_fun, sources)
assert_array_almost_equal(sources, ica.transform(m.T))
assert_equal(ica.mixing_.shape, (2, 2))
for fn in [np.tanh, "exp(-.5(x^2))"]:
ica = FastICA(fun=fn, algorithm=algo, random_state=0)
assert_raises(ValueError, ica.fit, m.T)
assert_raises(TypeError, FastICA(fun=moves.xrange(10)).fit, m.T)
