def test_parameters_sampler_replacement():
# raise error if n_iter too large
params = {'first': [0, 1], 'second': ['a', 'b', 'c']}
sampler = ParameterSampler(params, n_iter=7)
assert_raises(ValueError, list, sampler)
# degenerates to GridSearchCV if n_iter the same as grid_size
sampler = ParameterSampler(params, n_iter=6)
samples = list(sampler)
assert_equal(len(samples), 6)
for values in ParameterGrid(params):
assert_true(values in samples)
# test sampling without replacement in a large grid
params = {'a': range(10), 'b': range(10), 'c': range(10)}
sampler = ParameterSampler(params, n_iter=99, random_state=42)
samples = list(sampler)
assert_equal(len(samples), 99)
hashable_samples = ["a%db%dc%d" % (p['a'], p['b'], p['c'])
for p in samples]
assert_equal(len(set(hashable_samples)), 99)
# doesn't go into infinite loops
params_distribution = {'first': bernoulli(.5), 'second': ['a', 'b', 'c']}
sampler = ParameterSampler(params_distribution, n_iter=7)
samples = list(sampler)
assert_equal(len(samples), 7)
def test_radius_neighbors_classifier_when_no_neighbors():
""" Test radius-based classifier when no neighbors found.
In this case it should rise an informative exception """
X = np.array([[1.0, 1.0], [2.0, 2.0]])
y = np.array([1, 2])
radius = 0.1
z1 = np.array([[1.01, 1.01], [2.01, 2.01]]) # no outliers
z2 = np.array([[1.01, 1.01], [1.4, 1.4]]) # one outlier
weight_func = _weight_func
for outlier_label in [0, -1, None]:
for algorithm in ALGORITHMS:
for weights in ['uniform', 'distance', weight_func]:
rnc = neighbors.RadiusNeighborsClassifier
clf = rnc(radius=radius, weights=weights, algorithm=algorithm,
outlier_label=outlier_label)
clf.fit(X, y)
assert_array_equal(np.array([1, 2]),
clf.predict(z1))
if outlier_label is None:
assert_raises(ValueError, clf.predict, z2)
elif False:
assert_array_equal(np.array([1, outlier_label]),
clf.predict(z2))
def test_grid_search_precomputed_kernel():
"""Test that grid search works when the input features are given in the
form of a precomputed kernel matrix """
X_, y_ = make_classification(n_samples=200, n_features=100, random_state=0)
# compute the training kernel matrix corresponding to the linear kernel
K_train = np.dot(X_[:180], X_[:180].T)
y_train = y_[:180]
clf = SVC(kernel='precomputed')
cv = GridSearchCV(clf, {'C': [0.1, 1.0]})
cv.fit(K_train, y_train)
assert_true(cv.best_score_ >= 0)
# compute the test kernel matrix
K_test = np.dot(X_[180:], X_[:180].T)
y_test = y_[180:]
y_pred = cv.predict(K_test)
assert_true(np.mean(y_pred == y_test) >= 0)
# test error is raised when the precomputed kernel is not array-like
# or sparse
assert_raises(ValueError, cv.fit, K_train.tolist(), y_train)
def test_k_means_function():
# test calling the k_means function directly
# catch output
old_stdout = sys.stdout
sys.stdout = StringIO()
try:
cluster_centers, labels, inertia = k_means(X, n_clusters=n_clusters,
sample_weight=None,
verbose=True)
finally:
sys.stdout = old_stdout
centers = cluster_centers
assert_equal(centers.shape, (n_clusters, n_features))
labels = labels
assert_equal(np.unique(labels).shape[0], n_clusters)
# check that the labels assignment are perfect (up to a permutation)
assert_equal(v_measure_score(true_labels, labels), 1.0)
assert_greater(inertia, 0.0)
# check warning when centers are passed
assert_warns(RuntimeWarning, k_means, X, n_clusters=n_clusters,
sample_weight=None, init=centers)
# to many clusters desired
assert_raises(ValueError, k_means, X, n_clusters=X.shape[0] + 1,
sample_weight=None)
# kmeans for algorithm='elkan' raises TypeError on sparse matrix
assert_raise_message(TypeError, "algorithm='elkan' not supported for "
"sparse input X", k_means, X=X_csr, n_clusters=2,
sample_weight=None, algorithm="elkan")
def test_grid_search_precomputed_kernel_error_kernel_function():
"""Test that grid search returns an error when using a kernel_function"""
X_, y_ = make_classification(n_samples=200, n_features=100, random_state=0)
kernel_function = lambda x1, x2: np.dot(x1, x2.T)
clf = SVC(kernel=kernel_function)
cv = GridSearchCV(clf, {'C': [0.1, 1.0]})
assert_raises(ValueError, cv.fit, X_, y_)
def test_ovr_multilabel_predict_proba():
base_clf = MultinomialNB(alpha=1)
for au in (False, True):
X, Y = datasets.make_multilabel_classification(n_samples=100,
n_features=20,
n_classes=5,
n_labels=3,
length=50,
allow_unlabeled=au,
return_indicator=True,
random_state=0)
X_train, Y_train = X[:80], Y[:80]
X_test, Y_test = X[80:], Y[80:]
clf = OneVsRestClassifier(base_clf).fit(X_train, Y_train)
# decision function only estimator. Fails in current implementation.
decision_only = OneVsRestClassifier(svm.SVR()).fit(X_train, Y_train)
assert_raises(AttributeError, decision_only.predict_proba, X_test)
# Estimator with predict_proba disabled, depending on parameters.
decision_only = OneVsRestClassifier(svm.SVC(probability=False))
decision_only.fit(X_train, Y_train)
assert_raises(AttributeError, decision_only.predict_proba, X_test)
Y_pred = clf.predict(X_test)
Y_proba = clf.predict_proba(X_test)
# predict assigns a label if the probability that the
# sample has the label is greater than 0.5.
pred = Y_proba > .5
assert_array_equal(pred, Y_pred)
def test_k_means_function():
# test calling the k_means function directly
# catch output
old_stdout = sys.stdout
sys.stdout = StringIO()
try:
cluster_centers, labels, inertia = k_means(X, n_clusters=n_clusters,
verbose=True)
finally:
sys.stdout = old_stdout
centers = cluster_centers
assert_equal(centers.shape, (n_clusters, n_features))
labels = labels
assert_equal(np.unique(labels).shape[0], n_clusters)
# check that the labels assignment are perfect (up to a permutation)
assert_equal(v_measure_score(true_labels, labels), 1.0)
assert_greater(inertia, 0.0)
# check warning when centers are passed
assert_warns(RuntimeWarning, k_means, X, n_clusters=n_clusters,
init=centers)
# to many clusters desired
assert_raises(ValueError, k_means, X, n_clusters=X.shape[0] + 1)
def test_score_memmap():
# Ensure a scalar score of memmap type is accepted
iris = load_iris()
X, y = iris.data, iris.target
clf = MockClassifier()
tf = tempfile.NamedTemporaryFile(mode='wb', delete=False)
tf.write(b'Hello world!!!!!')
tf.close()
scores = np.memmap(tf.name, dtype=np.float64)
score = np.memmap(tf.name, shape=(), mode='r', dtype=np.float64)
try:
cross_val_score(clf, X, y, scoring=lambda est, X, y: score)
# non-scalar should still fail
assert_raises(ValueError, cross_val_score, clf, X, y,
scoring=lambda est, X, y: scores)
finally:
# Best effort to release the mmap file handles before deleting the
# backing file under Windows
scores, score = None, None
for _ in range(3):
try:
os.unlink(tf.name)
break
except WindowsError:
sleep(1.)
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 check_transformers_unfitted(name, Transformer):
X, y = _boston_subset()
with warnings.catch_warnings(record=True):
transformer = Transformer()
assert_raises(NotFittedError, transformer.transform, X)
def check_regressors_train(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 == 'PassiveAggressiveRegressor':
regressor.C = 0.01
# raises error on malformed input for fit
assert_raises(ValueError, regressor.fit, X, y[:-1])
# fit
if name in CROSS_DECOMPOSITION:
y_ = np.vstack([y, 2 * y + rnd.randint(2, size=len(y))])
y_ = y_.T
else:
y_ = y
set_random_state(regressor)
regressor.fit(X, y_)
regressor.fit(X.tolist(), y_.tolist())
regressor.predict(X)
# TODO: find out why PLS and CCA fail. RANSAC is random
# and furthermore assumes the presence of outliers, hence
# skipped
if name not in ('PLSCanonical', 'CCA', 'RANSACRegressor'):
print(regressor)
assert_greater(regressor.score(X, y_), 0.5)
def check_min_samples_leaf(name):
X, y = hastie_X, hastie_y
# Test if leaves contain more than leaf_count training examples
ForestEstimator = FOREST_ESTIMATORS[name]
# test boundary value
assert_raises(ValueError,
ForestEstimator(min_samples_leaf=-1).fit, X, y)
assert_raises(ValueError,
ForestEstimator(min_samples_leaf=0).fit, X, y)
est = ForestEstimator(min_samples_leaf=5, n_estimators=1, random_state=0)
est.fit(X, y)
out = est.estimators_[0].tree_.apply(X)
node_counts = np.bincount(out)
# drop inner nodes
leaf_count = node_counts[node_counts != 0]
assert_greater(np.min(leaf_count), 4,
"Failed with {0}".format(name))
est = ForestEstimator(min_samples_leaf=0.25, n_estimators=1,
random_state=0)
est.fit(X, y)
out = est.estimators_[0].tree_.apply(X)
node_counts = np.bincount(out)
# drop inner nodes
leaf_count = node_counts[node_counts != 0]
assert_greater(np.min(leaf_count), len(X) * 0.25 - 1,
"Failed with {0}".format(name))
def test_set_params():
# test nested estimator parameter setting
clf = Pipeline([("svc", SVC())])
# non-existing parameter in svc
assert_raises(ValueError, clf.set_params, svc__stupid_param=True)
# non-existing parameter of pipeline
assert_raises(ValueError, clf.set_params, svm__stupid_param=True)
def test_multioutput_number_of_output_differ():
y_true = np.array([[1, 0, 0, 1], [0, 1, 1, 1], [1, 1, 0, 1]])
y_pred = np.array([[0, 0], [1, 0], [0, 0]])
for name in MULTIOUTPUT_METRICS:
metric = ALL_METRICS[name]
assert_raises(ValueError, metric, y_true, y_pred)
def test_scikit_vs_scipy():
"""Test scikit linkage with full connectivity (i.e. unstructured) vs scipy
"""
n, p, k = 10, 5, 3
rng = np.random.RandomState(0)
# Not using a lil_matrix here, just to check that non sparse
# matrices are well handled
connectivity = np.ones((n, n))
for linkage in _TREE_BUILDERS.keys():
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 = hierarchy.linkage(X, method=linkage)
children_ = out[:, :2].astype(np.int)
children, _, n_leaves, _ = _TREE_BUILDERS[linkage](X, connectivity)
cut = _hc_cut(k, children, n_leaves)
cut_ = _hc_cut(k, children_, n_leaves)
assess_same_labelling(cut, cut_)
# Test error management in _hc_cut
assert_raises(ValueError, _hc_cut, n_leaves + 1, children, n_leaves)
def test_constant_size_multioutput_regressor():
random_state = np.random.RandomState(seed=1)
X = random_state.randn(10, 10)
y = random_state.randn(10, 5)
est = DummyRegressor(strategy='constant', constant=[1, 2, 3, 4])
assert_raises(ValueError, est.fit, X, y)
def test_cross_val_score():
clf = MockClassifier()
for a in range(-10, 10):
clf.a = a
# Smoke test
scores = cval.cross_val_score(clf, X, y)
assert_array_equal(scores, clf.score(X, y))
# test with multioutput y
scores = cval.cross_val_score(clf, X_sparse, X)
assert_array_equal(scores, clf.score(X_sparse, X))
scores = cval.cross_val_score(clf, X_sparse, y)
assert_array_equal(scores, clf.score(X_sparse, y))
# test with multioutput y
scores = cval.cross_val_score(clf, X_sparse, X)
assert_array_equal(scores, clf.score(X_sparse, X))
# test with X as list
clf = MockListClassifier()
scores = cval.cross_val_score(clf, X.tolist(), y)
assert_raises(ValueError, cval.cross_val_score, clf, X, y,
scoring="sklearn")
def test_nan():
# Test proper NaN handling.
# Regression test for Issue #252: fit used to go into an infinite loop.
Xnan = np.array(X, dtype=np.float64)
Xnan[0, 1] = np.nan
logistic = LogisticRegression(random_state=0)
assert_raises(ValueError, logistic.fit, Xnan, Y1)
def test_spectral_amg_mode():
# Test the amg mode of SpectralClustering
centers = np.array([
[0., 0., 0.],
[10., 10., 10.],
[20., 20., 20.],
])
X, true_labels = make_blobs(n_samples=100, centers=centers,
cluster_std=1., random_state=42)
D = pairwise_distances(X) # Distance matrix
S = np.max(D) - D # Similarity matrix
S = sparse.coo_matrix(S)
try:
from pyamg import smoothed_aggregation_solver
amg_loaded = True
except ImportError:
amg_loaded = False
if amg_loaded:
labels = spectral_clustering(S, n_clusters=len(centers),
random_state=0, mode="amg")
# We don't care too much that it's good, just that it *worked*.
# There does have to be some lower limit on the performance though.
assert_greater(np.mean(labels == true_labels), .3)
else:
assert_raises(ValueError, spectral_embedding, S,
n_components=len(centers), random_state=0, mode="amg")
def test_warm_start_smaller_n_estimators(Cls):
# Test if warm start with smaller n_estimators raises error
X, y = datasets.make_hastie_10_2(n_samples=100, random_state=1)
est = Cls(n_estimators=100, max_depth=1, warm_start=True)
est.fit(X, y)
est.set_params(n_estimators=99)
assert_raises(ValueError, est.fit, X, y)
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 check_boston(presort, loss, subsample):
# Check consistency on dataset boston house prices with least squares
# and least absolute deviation.
ones = np.ones(len(boston.target))
last_y_pred = None
for sample_weight in None, ones, 2 * ones:
clf = GradientBoostingRegressor(n_estimators=100,
loss=loss,
max_depth=4,
subsample=subsample,
min_samples_split=2,
random_state=1,
presort=presort)
assert_raises(ValueError, clf.predict, boston.data)
clf.fit(boston.data, boston.target,
sample_weight=sample_weight)
leaves = clf.apply(boston.data)
assert_equal(leaves.shape, (506, 100))
y_pred = clf.predict(boston.data)
mse = mean_squared_error(boston.target, y_pred)
assert_less(mse, 6.0)
if last_y_pred is not None:
assert_array_almost_equal(last_y_pred, y_pred)
last_y_pred = y_pred
def test_multilabel_binarizer_non_integer_labels():
tuple_classes = np.empty(3, dtype=object)
tuple_classes[:] = [(1,), (2,), (3,)]
inputs = [
([('2', '3'), ('1',), ('1', '2')], ['1', '2', '3']),
([('b', 'c'), ('a',), ('a', 'b')], ['a', 'b', 'c']),
([((2,), (3,)), ((1,),), ((1,), (2,))], tuple_classes),
]
indicator_mat = np.array([[0, 1, 1],
[1, 0, 0],
[1, 1, 0]])
for inp, classes in inputs:
# fit_transform()
mlb = MultiLabelBinarizer()
assert_array_equal(mlb.fit_transform(inp), indicator_mat)
assert_array_equal(mlb.classes_, classes)
assert_array_equal(mlb.inverse_transform(indicator_mat), inp)
# fit().transform()
mlb = MultiLabelBinarizer()
assert_array_equal(mlb.fit(inp).transform(inp), indicator_mat)
assert_array_equal(mlb.classes_, classes)
assert_array_equal(mlb.inverse_transform(indicator_mat), inp)
mlb = MultiLabelBinarizer()
assert_raises(TypeError, mlb.fit_transform, [({}), ({}, {'a': 'b'})])
def test_thresholded_scorers():
"""Test scorers that take thresholds."""
X, y = make_blobs(random_state=0, centers=2)
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)
clf = LogisticRegression(random_state=0)
clf.fit(X_train, y_train)
score1 = SCORERS['roc_auc'](clf, X_test, y_test)
score2 = roc_auc_score(y_test, clf.decision_function(X_test))
score3 = roc_auc_score(y_test, clf.predict_proba(X_test)[:, 1])
assert_almost_equal(score1, score2)
assert_almost_equal(score1, score3)
logscore = SCORERS['log_loss'](clf, X_test, y_test)
logloss = log_loss(y_test, clf.predict_proba(X_test))
assert_almost_equal(-logscore, logloss)
# same for an estimator without decision_function
clf = DecisionTreeClassifier()
clf.fit(X_train, y_train)
score1 = SCORERS['roc_auc'](clf, X_test, y_test)
score2 = roc_auc_score(y_test, clf.predict_proba(X_test)[:, 1])
assert_almost_equal(score1, score2)
# Test that an exception is raised on more than two classes
X, y = make_blobs(random_state=0, centers=3)
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)
clf.fit(X_train, y_train)
assert_raises(ValueError, SCORERS['roc_auc'], clf, X_test, y_test)
def test_partial_fit():
# Checks whether inserting array is consistent with fitted data.
# `partial_fit` method should set all attribute values correctly.
n_samples = 12
n_samples_partial_fit = 3
n_features = 2
rng = np.random.RandomState(42)
X = rng.rand(n_samples, n_features)
X_partial_fit = rng.rand(n_samples_partial_fit, n_features)
lshf = ignore_warnings(LSHForest, category=DeprecationWarning)()
# Test unfitted estimator
ignore_warnings(lshf.partial_fit)(X)
assert_array_equal(X, lshf._fit_X)
ignore_warnings(lshf.fit)(X)
# Insert wrong dimension
assert_raises(ValueError, lshf.partial_fit,
np.random.randn(n_samples_partial_fit, n_features - 1))
ignore_warnings(lshf.partial_fit)(X_partial_fit)
# size of _input_array = samples + 1 after insertion
assert_equal(lshf._fit_X.shape[0],
n_samples + n_samples_partial_fit)
# size of original_indices_[1] = samples + 1
assert_equal(len(lshf.original_indices_[0]),
n_samples + n_samples_partial_fit)
# size of trees_[1] = samples + 1
assert_equal(len(lshf.trees_[1]),
n_samples + n_samples_partial_fit)
def test_kernel_density_sampling(n_samples=100, n_features=3):
rng = np.random.RandomState(0)
X = rng.randn(n_samples, n_features)
bandwidth = 0.2
for kernel in ['gaussian', 'tophat']:
# draw a tophat sample
kde = KernelDensity(bandwidth, kernel=kernel).fit(X)
samp = kde.sample(100)
assert_equal(X.shape, samp.shape)
# check that samples are in the right range
nbrs = NearestNeighbors(n_neighbors=1).fit(X)
dist, ind = nbrs.kneighbors(X, return_distance=True)
if kernel == 'tophat':
assert np.all(dist < bandwidth)
elif kernel == 'gaussian':
# 5 standard deviations is safe for 100 samples, but there's a
# very small chance this test could fail.
assert np.all(dist < 5 * bandwidth)
# check unsupported kernels
for kernel in ['epanechnikov', 'exponential', 'linear', 'cosine']:
kde = KernelDensity(bandwidth, kernel=kernel).fit(X)
assert_raises(NotImplementedError, kde.sample, 100)
# non-regression test: used to return a scalar
X = rng.randn(4, 1)
kde = KernelDensity(kernel="gaussian").fit(X)
assert_equal(kde.sample().shape, (1, 1))
def test_pca_randomized_solver():
# PCA on dense arrays
X = iris.data
# Loop excluding the 0, invalid for randomized
for n_comp in np.arange(1, X.shape[1]):
pca = PCA(n_components=n_comp, svd_solver='randomized', random_state=0)
X_r = pca.fit(X).transform(X)
np.testing.assert_equal(X_r.shape[1], n_comp)
X_r2 = pca.fit_transform(X)
assert_array_almost_equal(X_r, X_r2)
X_r = pca.transform(X)
assert_array_almost_equal(X_r, X_r2)
# Test get_covariance and get_precision
cov = pca.get_covariance()
precision = pca.get_precision()
assert_array_almost_equal(np.dot(cov, precision),
np.eye(X.shape[1]), 12)
pca = PCA(n_components=0, svd_solver='randomized', random_state=0)
assert_raises(ValueError, pca.fit, X)
pca = PCA(n_components=0, svd_solver='randomized', random_state=0)
assert_raises(ValueError, pca.fit, X)
# Check internal state
assert_equal(pca.n_components,
PCA(n_components=0,
svd_solver='randomized', random_state=0).n_components)
assert_equal(pca.svd_solver,
PCA(n_components=0,
svd_solver='randomized', random_state=0).svd_solver)
def test_zero_estimator_clf():
# Test if ZeroEstimator works for classification.
X = iris.data
y = np.array(iris.target)
est = GradientBoostingClassifier(n_estimators=20, max_depth=1,
random_state=1, init=ZeroEstimator())
est.fit(X, y)
assert_greater(est.score(X, y), 0.96)
est = GradientBoostingClassifier(n_estimators=20, max_depth=1,
random_state=1, init='zero')
est.fit(X, y)
assert_greater(est.score(X, y), 0.96)
# binary clf
mask = y != 0
y[mask] = 1
y[~mask] = 0
est = GradientBoostingClassifier(n_estimators=20, max_depth=1,
random_state=1, init='zero')
est.fit(X, y)
assert_greater(est.score(X, y), 0.96)
est = GradientBoostingClassifier(n_estimators=20, max_depth=1,
random_state=1, init='foobar')
assert_raises(ValueError, est.fit, X, y)
def test_ward_clustering():
"""
Check that we obtain the correct number of clusters with Ward clustering.
"""
rnd = np.random.RandomState(0)
mask = np.ones([10, 10], dtype=np.bool)
X = rnd.randn(100, 50)
connectivity = grid_to_graph(*mask.shape)
clustering = Ward(n_clusters=10, connectivity=connectivity)
clustering.fit(X)
# test caching
clustering = Ward(n_clusters=10, connectivity=connectivity,
memory=mkdtemp())
clustering.fit(X)
labels = clustering.labels_
assert_true(np.size(np.unique(labels)) == 10)
# Turn caching off now
clustering = Ward(n_clusters=10, connectivity=connectivity)
# Check that we obtain the same solution with early-stopping of the
# tree building
clustering.compute_full_tree = False
clustering.fit(X)
np.testing.assert_array_equal(clustering.labels_, labels)
clustering.connectivity = None
clustering.fit(X)
assert_true(np.size(np.unique(clustering.labels_)) == 10)
# Check that we raise a TypeError on dense matrices
clustering = Ward(n_clusters=10,
connectivity=connectivity.todense())
assert_raises(TypeError, clustering.fit, X)
clustering = Ward(n_clusters=10,
connectivity=sparse.lil_matrix(
connectivity.todense()[:10, :10]))
assert_raises(ValueError, clustering.fit, X)
def test_staged_predict_proba():
# Test whether staged predict proba eventually gives
# the same prediction.
X, y = datasets.make_hastie_10_2(n_samples=1200,
random_state=1)
X_train, y_train = X[:200], y[:200]
X_test, y_test = X[200:], y[200:]
clf = GradientBoostingClassifier(n_estimators=20)
# test raise NotFittedError if not fitted
assert_raises(NotFittedError, lambda X: np.fromiter(
clf.staged_predict_proba(X), dtype=np.float64), X_test)
clf.fit(X_train, y_train)
# test if prediction for last stage equals ``predict``
for y_pred in clf.staged_predict(X_test):
assert_equal(y_test.shape, y_pred.shape)
assert_array_equal(clf.predict(X_test), y_pred)
# test if prediction for last stage equals ``predict_proba``
for staged_proba in clf.staged_predict_proba(X_test):
assert_equal(y_test.shape[0], staged_proba.shape[0])
assert_equal(2, staged_proba.shape[1])
assert_array_almost_equal(clf.predict_proba(X_test), staged_proba)
def test_bad_input():
# Test that it gives proper exception on deficient input
# impossible value of C
assert_raises(ValueError, svm.SVC(C=-1).fit, X, Y)
# impossible value of nu
clf = svm.NuSVC(nu=0.0)
assert_raises(ValueError, clf.fit, X, Y)
Y2 = Y[:-1] # wrong dimensions for labels
assert_raises(ValueError, clf.fit, X, Y2)
# Test with arrays that are non-contiguous.
for clf in (svm.SVC(), svm.LinearSVC(random_state=0)):
Xf = np.asfortranarray(X)
assert not Xf.flags['C_CONTIGUOUS']
yf = np.ascontiguousarray(np.tile(Y, (2, 1)).T)
yf = yf[:, -1]
assert not yf.flags['F_CONTIGUOUS']
assert not yf.flags['C_CONTIGUOUS']
clf.fit(Xf, yf)
assert_array_equal(clf.predict(T), true_result)
# error for precomputed kernelsx
clf = svm.SVC(kernel='precomputed')
assert_raises(ValueError, clf.fit, X, Y)
# sample_weight bad dimensions
clf = svm.SVC()
assert_raises(ValueError, clf.fit, X, Y, sample_weight=range(len(X) - 1))
# predict with sparse input when trained with dense
clf = svm.SVC().fit(X, Y)
assert_raises(ValueError, clf.predict, sparse.lil_matrix(X))
Xt = np.array(X).T
clf.fit(np.dot(X, Xt), Y)
assert_raises(ValueError, clf.predict, X)
clf = svm.SVC()
clf.fit(X, Y)
assert_raises(ValueError, clf.predict, Xt)
def test_stratified_shuffle_split_init():
y = np.asarray([0, 1, 1, 1, 2, 2, 2])
# Check that error is raised if there is a class with only one sample
assert_raises(ValueError, cval.StratifiedShuffleSplit, y, 3, 0.2)
# Check that error is raised if the test set size is smaller than n_classes
assert_raises(ValueError, cval.StratifiedShuffleSplit, y, 3, 2)
# Check that error is raised if the train set size is smaller than
# n_classes
assert_raises(ValueError, cval.StratifiedShuffleSplit, y, 3, 3, 2)
y = np.asarray([0, 0, 0, 1, 1, 1, 2, 2, 2])
# Check that errors are raised if there is not enough samples
assert_raises(ValueError, cval.StratifiedShuffleSplit, y, 3, 0.5, 0.6)
assert_raises(ValueError, cval.StratifiedShuffleSplit, y, 3, 8, 0.6)
assert_raises(ValueError, cval.StratifiedShuffleSplit, y, 3, 0.6, 8)
# Train size or test size too small
assert_raises(ValueError, cval.StratifiedShuffleSplit, y, train_size=2)
assert_raises(ValueError, cval.StratifiedShuffleSplit, y, test_size=2)
def check_label_kfold(shuffle):
rng = np.random.RandomState(0)
# Parameters of the test
n_labels = 15
n_samples = 1000
n_folds = 5
# Construct the test data
tolerance = 0.05 * n_samples # 5 percent error allowed
labels = rng.randint(0, n_labels, n_samples)
folds = cval.LabelKFold(labels,
n_folds=n_folds,
shuffle=shuffle,
random_state=rng).idxs
ideal_n_labels_per_fold = n_samples // n_folds
# Check that folds have approximately the same size
assert_equal(len(folds), len(labels))
for i in np.unique(folds):
assert_greater_equal(tolerance,
abs(sum(folds == i) - ideal_n_labels_per_fold))
# Check that each label appears only in 1 fold
for label in np.unique(labels):
assert_equal(len(np.unique(folds[labels == label])), 1)
# Check that no label is on both sides of the split
labels = np.asarray(labels, dtype=object)
for train, test in cval.LabelKFold(labels,
n_folds=n_folds,
shuffle=shuffle,
random_state=rng):
assert_equal(len(np.intersect1d(labels[train], labels[test])), 0)
# Construct the test data
labels = [
'Albert', 'Jean', 'Bertrand', 'Michel', 'Jean', 'Francis', 'Robert',
'Michel', 'Rachel', 'Lois', 'Michelle', 'Bernard', 'Marion', 'Laura',
'Jean', 'Rachel', 'Franck', 'John', 'Gael', 'Anna', 'Alix', 'Robert',
'Marion', 'David', 'Tony', 'Abel', 'Becky', 'Madmood', 'Cary', 'Mary',
'Alexandre', 'David', 'Francis', 'Barack', 'Abdoul', 'Rasha', 'Xi',
'Silvia'
]
labels = np.asarray(labels, dtype=object)
n_labels = len(np.unique(labels))
n_samples = len(labels)
n_folds = 5
tolerance = 0.05 * n_samples # 5 percent error allowed
folds = cval.LabelKFold(labels,
n_folds=n_folds,
shuffle=shuffle,
random_state=rng).idxs
ideal_n_labels_per_fold = n_samples // n_folds
# Check that folds have approximately the same size
assert_equal(len(folds), len(labels))
for i in np.unique(folds):
assert_greater_equal(tolerance,
abs(sum(folds == i) - ideal_n_labels_per_fold))
# Check that each label appears only in 1 fold
for label in np.unique(labels):
assert_equal(len(np.unique(folds[labels == label])), 1)
# Check that no label is on both sides of the split
for train, test in cval.LabelKFold(labels,
n_folds=n_folds,
shuffle=shuffle,
random_state=rng):
assert_equal(len(np.intersect1d(labels[train], labels[test])), 0)
# Should fail if there are more folds than labels
labels = np.array([1, 1, 1, 2, 2])
assert_raises(ValueError, cval.LabelKFold, labels, n_folds=3)
def test_kfold_valueerrors():
# Check that errors are raised if there is not enough samples
assert_raises(ValueError, cval.KFold, 3, 4)
# Check that a warning is raised if the least populated class has too few
# members.
y = [3, 3, -1, -1, 2]
cv = assert_warns_message(Warning, "The least populated class",
cval.StratifiedKFold, y, 3)
# Check that despite the warning the folds are still computed even
# though all the classes are not necessarily represented at on each
# side of the split at each split
check_cv_coverage(cv, expected_n_iter=3, n_samples=len(y))
# Error when number of folds is <= 1
assert_raises(ValueError, cval.KFold, 2, 0)
assert_raises(ValueError, cval.KFold, 2, 1)
assert_raises(ValueError, cval.StratifiedKFold, y, 0)
assert_raises(ValueError, cval.StratifiedKFold, y, 1)
# When n is not integer:
assert_raises(ValueError, cval.KFold, 2.5, 2)
# When n_folds is not integer:
assert_raises(ValueError, cval.KFold, 5, 1.5)
assert_raises(ValueError, cval.StratifiedKFold, y, 1.5)
def test_shufflesplit_errors():
assert_raises(ValueError, cval.ShuffleSplit, 10, test_size=2.0)
assert_raises(ValueError, cval.ShuffleSplit, 10, test_size=1.0)
assert_raises(ValueError,
cval.ShuffleSplit,
10,
test_size=0.1,
train_size=0.95)
assert_raises(ValueError, cval.ShuffleSplit, 10, test_size=11)
assert_raises(ValueError, cval.ShuffleSplit, 10, test_size=10)
assert_raises(ValueError, cval.ShuffleSplit, 10, test_size=8, train_size=3)
assert_raises(ValueError, cval.ShuffleSplit, 10, train_size=1j)
assert_raises(ValueError,
cval.ShuffleSplit,
10,
test_size=None,
train_size=None)
def test_precomputed():
# SVC with a precomputed kernel.
# We test it with a toy dataset and with iris.
clf = svm.SVC(kernel='precomputed')
# Gram matrix for train data (square matrix)
# (we use just a linear kernel)
K = np.dot(X, np.array(X).T)
clf.fit(K, Y)
# Gram matrix for test data (rectangular matrix)
KT = np.dot(T, np.array(X).T)
pred = clf.predict(KT)
assert_raises(ValueError, clf.predict, KT.T)
assert_array_equal(clf.dual_coef_, [[-0.25, .25]])
assert_array_equal(clf.support_, [1, 3])
assert_array_equal(clf.intercept_, [0])
assert_array_almost_equal(clf.support_, [1, 3])
assert_array_equal(pred, true_result)
# Gram matrix for test data but compute KT[i,j]
# for support vectors j only.
KT = np.zeros_like(KT)
for i in range(len(T)):
for j in clf.support_:
KT[i, j] = np.dot(T[i], X[j])
pred = clf.predict(KT)
assert_array_equal(pred, true_result)
# same as before, but using a callable function instead of the kernel
# matrix. kernel is just a linear kernel
kfunc = lambda x, y: np.dot(x, y.T)
clf = svm.SVC(kernel=kfunc)
clf.fit(X, Y)
pred = clf.predict(T)
assert_array_equal(clf.dual_coef_, [[-0.25, .25]])
assert_array_equal(clf.intercept_, [0])
assert_array_almost_equal(clf.support_, [1, 3])
assert_array_equal(pred, true_result)
# test a precomputed kernel with the iris dataset
# and check parameters against a linear SVC
clf = svm.SVC(kernel='precomputed')
clf2 = svm.SVC(kernel='linear')
K = np.dot(iris.data, iris.data.T)
clf.fit(K, iris.target)
clf2.fit(iris.data, iris.target)
pred = clf.predict(K)
assert_array_almost_equal(clf.support_, clf2.support_)
assert_array_almost_equal(clf.dual_coef_, clf2.dual_coef_)
assert_array_almost_equal(clf.intercept_, clf2.intercept_)
assert_almost_equal(np.mean(pred == iris.target), .99, decimal=2)
# Gram matrix for test data but compute KT[i,j]
# for support vectors j only.
K = np.zeros_like(K)
for i in range(len(iris.data)):
for j in clf.support_:
K[i, j] = np.dot(iris.data[i], iris.data[j])
pred = clf.predict(K)
assert_almost_equal(np.mean(pred == iris.target), .99, decimal=2)
clf = svm.SVC(kernel=kfunc)
clf.fit(iris.data, iris.target)
assert_almost_equal(np.mean(pred == iris.target), .99, decimal=2)
def test_svc_bad_kernel():
svc = svm.SVC(kernel=lambda x, y: x)
assert_raises(ValueError, svc.fit, X, Y)
def check_unfitted_feature_importances(name):
assert_raises(ValueError, getattr, FOREST_ESTIMATORS[name](random_state=0),
"feature_importances_")
def test_prediction_proba_parameter(self):
with assert_raises(ValueError):
self.clf.predict_proba(self.X_test, method='something')
def test_check_array():
# accept_sparse == False
# raise error on sparse inputs
X = [[1, 2], [3, 4]]
X_csr = sp.csr_matrix(X)
assert_raises(TypeError, check_array, X_csr)
# ensure_2d=False
X_array = check_array([0, 1, 2], ensure_2d=False)
assert X_array.ndim == 1
# ensure_2d=True with 1d array
assert_raise_message(ValueError,
'Expected 2D array, got 1D array instead',
check_array, [0, 1, 2],
ensure_2d=True)
# ensure_2d=True with scalar array
assert_raise_message(ValueError,
'Expected 2D array, got scalar array instead',
check_array,
10,
ensure_2d=True)
# don't allow ndim > 3
X_ndim = np.arange(8).reshape(2, 2, 2)
assert_raises(ValueError, check_array, X_ndim)
check_array(X_ndim, allow_nd=True) # doesn't raise
# dtype and order enforcement.
X_C = np.arange(4).reshape(2, 2).copy("C")
X_F = X_C.copy("F")
X_int = X_C.astype(np.int)
X_float = X_C.astype(np.float)
Xs = [X_C, X_F, X_int, X_float]
dtypes = [np.int32, np.int, np.float, np.float32, None, np.bool, object]
orders = ['C', 'F', None]
copys = [True, False]
for X, dtype, order, copy in product(Xs, dtypes, orders, copys):
X_checked = check_array(X, dtype=dtype, order=order, copy=copy)
if dtype is not None:
assert X_checked.dtype == dtype
else:
assert X_checked.dtype == X.dtype
if order == 'C':
assert X_checked.flags['C_CONTIGUOUS']
assert not X_checked.flags['F_CONTIGUOUS']
elif order == 'F':
assert X_checked.flags['F_CONTIGUOUS']
assert not X_checked.flags['C_CONTIGUOUS']
if copy:
assert X is not X_checked
else:
# doesn't copy if it was already good
if (X.dtype == X_checked.dtype and X_checked.flags['C_CONTIGUOUS']
== X.flags['C_CONTIGUOUS']
and X_checked.flags['F_CONTIGUOUS']
== X.flags['F_CONTIGUOUS']):
assert X is X_checked
# allowed sparse != None
X_csc = sp.csc_matrix(X_C)
X_coo = X_csc.tocoo()
X_dok = X_csc.todok()
X_int = X_csc.astype(np.int)
X_float = X_csc.astype(np.float)
Xs = [X_csc, X_coo, X_dok, X_int, X_float]
accept_sparses = [['csr', 'coo'], ['coo', 'dok']]
for X, dtype, accept_sparse, copy in product(Xs, dtypes, accept_sparses,
copys):
with warnings.catch_warnings(record=True) as w:
X_checked = check_array(X,
dtype=dtype,
accept_sparse=accept_sparse,
copy=copy)
if (dtype is object or sp.isspmatrix_dok(X)) and len(w):
message = str(w[0].message)
messages = [
"object dtype is not supported by sparse matrices",
"Can't check dok sparse matrix for nan or inf."
]
assert message in messages
else:
assert len(w) == 0
if dtype is not None:
assert X_checked.dtype == dtype
else:
assert X_checked.dtype == X.dtype
if X.format in accept_sparse:
# no change if allowed
assert X.format == X_checked.format
else:
# got converted
assert X_checked.format == accept_sparse[0]
if copy:
assert X is not X_checked
else:
# doesn't copy if it was already good
if X.dtype == X_checked.dtype and X.format == X_checked.format:
assert X is X_checked
# other input formats
# convert lists to arrays
X_dense = check_array([[1, 2], [3, 4]])
assert isinstance(X_dense, np.ndarray)
# raise on too deep lists
assert_raises(ValueError, check_array, X_ndim.tolist())
check_array(X_ndim.tolist(), allow_nd=True) # doesn't raise
# convert weird stuff to arrays
X_no_array = _NotAnArray(X_dense)
result = check_array(X_no_array)
assert isinstance(result, np.ndarray)
# deprecation warning if string-like array with dtype="numeric"
expected_warn_regex = r"converted to decimal numbers if dtype='numeric'"
X_str = [['11', '12'], ['13', 'xx']]
for X in [X_str, np.array(X_str, dtype='U'), np.array(X_str, dtype='S')]:
with pytest.warns(FutureWarning, match=expected_warn_regex):
check_array(X, dtype="numeric")
# deprecation warning if byte-like array with dtype="numeric"
X_bytes = [[b'a', b'b'], [b'c', b'd']]
for X in [X_bytes, np.array(X_bytes, dtype='V1')]:
with pytest.warns(FutureWarning, match=expected_warn_regex):
check_array(X, dtype="numeric")
def test_pipeline_init():
# Test the various init parameters of the pipeline.
assert_raises(TypeError, Pipeline)
# Check that we can't instantiate pipelines with objects without fit
# method
assert_raises_regex(
TypeError, 'Last step of Pipeline should implement fit. '
'.*NoFit.*', Pipeline, [('clf', NoFit())])
# Smoke test with only an estimator
clf = NoTrans()
pipe = Pipeline([('svc', clf)])
assert_equal(
pipe.get_params(deep=True),
dict(svc__a=None, svc__b=None, svc=clf, **pipe.get_params(deep=False)))
# Check that params are set
pipe.set_params(svc__a=0.1)
assert_equal(clf.a, 0.1)
assert_equal(clf.b, None)
# Smoke test the repr:
repr(pipe)
# Test with two objects
clf = SVC()
filter1 = SelectKBest(f_classif)
pipe = Pipeline([('anova', filter1), ('svc', clf)])
# Check that we can't instantiate with non-transformers on the way
# Note that NoTrans implements fit, but not transform
assert_raises_regex(
TypeError, 'All intermediate steps should be transformers'
'.*\\bNoTrans\\b.*', Pipeline, [('t', NoTrans()), ('svc', clf)])
# Check that params are set
pipe.set_params(svc__C=0.1)
assert_equal(clf.C, 0.1)
# Smoke test the repr:
repr(pipe)
# Check that params are not set when naming them wrong
assert_raises(ValueError, pipe.set_params, anova__C=0.1)
# Test clone
pipe2 = clone(pipe)
assert_false(pipe.named_steps['svc'] is pipe2.named_steps['svc'])
# Check that apart from estimators, the parameters are the same
params = pipe.get_params(deep=True)
params2 = pipe2.get_params(deep=True)
for x in pipe.get_params(deep=False):
params.pop(x)
for x in pipe2.get_params(deep=False):
params2.pop(x)
# Remove estimators that where copied
params.pop('svc')
params.pop('anova')
params2.pop('svc')
params2.pop('anova')
assert_equal(params, params2)
def test_iforest_error():
"""Test that it gives proper exception on deficient input."""
X = iris.data
# Test max_samples
assert_raises(ValueError, IsolationForest(max_samples=-1).fit, X)
assert_raises(ValueError, IsolationForest(max_samples=0.0).fit, X)
assert_raises(ValueError, IsolationForest(max_samples=2.0).fit, X)
# The dataset has less than 256 samples, explicitly setting
# max_samples > n_samples should result in a warning. If not set
# explicitly there should be no warning
assert_warns_message(
UserWarning, "max_samples will be set to n_samples for estimation",
IsolationForest(max_samples=1000).fit, X)
assert_no_warnings(IsolationForest(max_samples='auto').fit, X)
assert_no_warnings(IsolationForest(max_samples=np.int64(2)).fit, X)
assert_raises(ValueError, IsolationForest(max_samples='foobar').fit, X)
assert_raises(ValueError, IsolationForest(max_samples=1.5).fit, X)
# test X_test n_features match X_train one:
assert_raises(ValueError, IsolationForest().fit(X).predict, X[:, 1:])
def test_neighbors_badargs():
"""Test bad argument values: these should all raise ValueErrors"""
assert_raises(ValueError,
neighbors.NearestNeighbors,
algorithm='blah')
X = rng.random_sample((10, 2))
Xsparse = csr_matrix(X)
y = np.ones(10)
for cls in (neighbors.KNeighborsClassifier,
neighbors.RadiusNeighborsClassifier,
neighbors.KNeighborsRegressor,
neighbors.RadiusNeighborsRegressor):
assert_raises(ValueError,
cls,
weights='blah')
assert_raises(ValueError,
cls, p=-1)
assert_raises(ValueError,
cls, algorithm='blah')
nbrs = cls(algorithm='ball_tree', metric='haversine')
assert_raises(ValueError,
nbrs.predict,
X)
assert_raises(ValueError,
ignore_warnings(nbrs.fit),
Xsparse, y)
nbrs = cls()
assert_raises(ValueError,
nbrs.fit,
np.ones((0, 2)), np.ones(0))
assert_raises(ValueError,
nbrs.fit,
X[:, :, None], y)
nbrs.fit(X, y)
assert_raises(ValueError,
nbrs.predict,
[])
nbrs = neighbors.NearestNeighbors().fit(X)
assert_raises(ValueError,
nbrs.kneighbors_graph,
X, mode='blah')
assert_raises(ValueError,
nbrs.radius_neighbors_graph,
X, mode='blah')
def test_weighted_dbscan():
# ensure sample_weight is validated
assert_raises(ValueError, dbscan, [[0], [1]], sample_weight=[2])
assert_raises(ValueError, dbscan, [[0], [1]], sample_weight=[2, 3, 4])
# ensure sample_weight has an effect
assert_array_equal([],
dbscan([[0], [1]], sample_weight=None,
min_samples=6)[0])
assert_array_equal([],
dbscan([[0], [1]], sample_weight=[5, 5],
min_samples=6)[0])
assert_array_equal([0],
dbscan([[0], [1]], sample_weight=[6, 5],
min_samples=6)[0])
assert_array_equal([0, 1],
dbscan([[0], [1]], sample_weight=[6, 6],
min_samples=6)[0])
# points within eps of each other:
assert_array_equal([0, 1],
dbscan([[0], [1]],
eps=1.5,
sample_weight=[5, 1],
min_samples=6)[0])
# and effect of non-positive and non-integer sample_weight:
assert_array_equal([],
dbscan([[0], [1]],
sample_weight=[5, 0],
eps=1.5,
min_samples=6)[0])
assert_array_equal([0, 1],
dbscan([[0], [1]],
sample_weight=[5.9, 0.1],
eps=1.5,
min_samples=6)[0])
assert_array_equal([0, 1],
dbscan([[0], [1]],
sample_weight=[6, 0],
eps=1.5,
min_samples=6)[0])
assert_array_equal([],
dbscan([[0], [1]],
sample_weight=[6, -1],
eps=1.5,
min_samples=6)[0])
# for non-negative sample_weight, cores should be identical to repetition
rng = np.random.RandomState(42)
sample_weight = rng.randint(0, 5, X.shape[0])
core1, label1 = dbscan(X, sample_weight=sample_weight)
assert_equal(len(label1), len(X))
X_repeated = np.repeat(X, sample_weight, axis=0)
core_repeated, label_repeated = dbscan(X_repeated)
core_repeated_mask = np.zeros(X_repeated.shape[0], dtype=bool)
core_repeated_mask[core_repeated] = True
core_mask = np.zeros(X.shape[0], dtype=bool)
core_mask[core1] = True
assert_array_equal(np.repeat(core_mask, sample_weight), core_repeated_mask)
# sample_weight should work with precomputed distance matrix
D = pairwise_distances(X)
core3, label3 = dbscan(D,
sample_weight=sample_weight,
metric='precomputed')
assert_array_equal(core1, core3)
assert_array_equal(label1, label3)
# sample_weight should work with estimator
est = DBSCAN().fit(X, sample_weight=sample_weight)
core4 = est.core_sample_indices_
label4 = est.labels_
assert_array_equal(core1, core4)
assert_array_equal(label1, label4)
est = DBSCAN()
label5 = est.fit_predict(X, sample_weight=sample_weight)
core5 = est.core_sample_indices_
assert_array_equal(core1, core5)
assert_array_equal(label1, label5)
assert_array_equal(label1, est.labels_)
def test_classifiers_train():
# test if classifiers do something sensible on training set
# also test all shapes / shape errors
classifiers = all_estimators(type_filter='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)]:
# do it once with binary, once with multiclass
classes = np.unique(y)
n_classes = len(classes)
n_samples, n_features = X.shape
for name, Classifier in classifiers:
if name in dont_test:
continue
if name in ['MultinomialNB', 'BernoulliNB']:
# TODO also test these!
continue
# catch deprecation warnings
with warnings.catch_warnings(record=True):
classifier = Classifier()
# raises error on malformed input for fit
assert_raises(ValueError, classifier.fit, X, y[:-1])
# fit
classifier.fit(X, y)
assert_true(hasattr(classifier, "classes_"))
y_pred = classifier.predict(X)
assert_equal(y_pred.shape, (n_samples, ))
# training set performance
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.ravel().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"):
try:
# 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)
except NotImplementedError:
pass
def test_multioutput_enetcv_error():
X = np.random.randn(10, 2)
y = np.random.randn(10, 2)
clf = ElasticNetCV()
assert_raises(ValueError, clf.fit, X, y)
def test_hdbscan_no_centroid_medoid_for_noise():
clusterer = HDBSCAN().fit(X)
assert_raises(ValueError, clusterer.weighted_cluster_centroid, -1)
assert_raises(ValueError, clusterer.weighted_cluster_medoid, -1)
def test_fast_dot():
# Check fast dot blas wrapper function
if fast_dot is np.dot:
return
rng = np.random.RandomState(42)
A = rng.random_sample([2, 10])
B = rng.random_sample([2, 10])
try:
linalg.get_blas_funcs(['gemm'])[0]
has_blas = True
except (AttributeError, ValueError):
has_blas = False
if has_blas:
# Test _fast_dot for invalid input.
# Maltyped data.
for dt1, dt2 in [['f8', 'f4'], ['i4', 'i4']]:
assert_raises(ValueError, _fast_dot, A.astype(dt1),
B.astype(dt2).T)
# Malformed data.
# ndim == 0
E = np.empty(0)
assert_raises(ValueError, _fast_dot, E, E)
# ndim == 1
assert_raises(ValueError, _fast_dot, A, A[0])
# ndim > 2
assert_raises(ValueError, _fast_dot, A.T, np.array([A, A]))
# min(shape) == 1
assert_raises(ValueError, _fast_dot, A, A[0, :][None, :])
# test for matrix mismatch error
assert_raises(ValueError, _fast_dot, A, A)
# Test cov-like use case + dtypes.
for dtype in ['f8', 'f4']:
A = A.astype(dtype)
B = B.astype(dtype)
# col < row
C = np.dot(A.T, A)
C_ = fast_dot(A.T, A)
assert_almost_equal(C, C_, decimal=5)
C = np.dot(A.T, B)
C_ = fast_dot(A.T, B)
assert_almost_equal(C, C_, decimal=5)
C = np.dot(A, B.T)
C_ = fast_dot(A, B.T)
assert_almost_equal(C, C_, decimal=5)
# Test square matrix * rectangular use case.
A = rng.random_sample([2, 2])
for dtype in ['f8', 'f4']:
A = A.astype(dtype)
B = B.astype(dtype)
C = np.dot(A, B)
C_ = fast_dot(A, B)
assert_almost_equal(C, C_, decimal=5)
C = np.dot(A.T, B)
C_ = fast_dot(A.T, B)
assert_almost_equal(C, C_, decimal=5)
if has_blas:
for x in [np.array([[d] * 10] * 2) for d in [np.inf, np.nan]]:
assert_raises(ValueError, _fast_dot, x, x.T)
def test_save_filepath_condition():
r = BaseRecommender()
invalid_filepath = 'no suffix'
assert_raises(ValueError, r.save, invalid_filepath)
def test_transformers():
# test if transformers do something sensible on training set
# also test all shapes / shape errors
transformers = all_estimators(type_filter='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()
succeeded = True
for name, Transformer in transformers:
if name in dont_test:
continue
# these don't actually fit the data:
if name in ['AdditiveChi2Sampler', 'Binarizer', 'Normalizer']:
continue
# catch deprecation warnings
with warnings.catch_warnings(record=True):
transformer = Transformer()
set_random_state(transformer)
if hasattr(transformer, 'compute_importances'):
transformer.compute_importances = True
if name == 'SelectKBest':
# SelectKBest has a default of k=10
# which is more feature than we have.
transformer.k = 1
elif name in ['GaussianRandomProjection', 'SparseRandomProjection']:
# Due to the jl lemma and very few samples, the number
# of components of the random matrix projection will be greater
# than the number of features.
# So we impose a smaller number (avoid "auto" mode)
transformer.n_components = 1
elif name == "MiniBatchDictionaryLearning":
transformer.set_params(n_iter=5) # default = 1000
elif name == "KernelPCA":
transformer.remove_zero_eig = False
# fit
if name in ('_PLS', 'PLSCanonical', 'PLSRegression', 'CCA', 'PLSSVD'):
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
try:
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)
except Exception as e:
print(transformer)
print(e)
print()
succeeded = False
continue
if hasattr(transformer, 'transform'):
if name in ('_PLS', 'PLSCanonical', 'PLSRegression', 'CCA',
'PLSSVD'):
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 not correct in %s" % Transformer)
assert_array_almost_equal(
x_pred3, x_pred2, 2,
"fit_transform not correct in %s" % Transformer)
else:
assert_array_almost_equal(
X_pred, X_pred2, 2,
"fit_transform not correct in %s" % Transformer)
assert_array_almost_equal(
X_pred3, X_pred2, 2,
"fit_transform not correct in %s" % Transformer)
# raises error on malformed input for transform
assert_raises(ValueError, transformer.transform, X.T)
assert_true(succeeded)
def test_agglomerative_clustering():
"""
Check that we obtain the correct number of clusters with
agglomerative clustering.
"""
rnd = np.random.RandomState(0)
mask = np.ones([10, 10], dtype=np.bool)
n_samples = 100
X = rnd.randn(n_samples, 50)
connectivity = grid_to_graph(*mask.shape)
for linkage in ("ward", "complete", "average"):
clustering = AgglomerativeClustering(n_clusters=10,
connectivity=connectivity,
linkage=linkage)
clustering.fit(X)
# test caching
clustering = AgglomerativeClustering(n_clusters=10,
connectivity=connectivity,
memory=mkdtemp(),
linkage=linkage)
clustering.fit(X)
labels = clustering.labels_
assert_true(np.size(np.unique(labels)) == 10)
# Turn caching off now
clustering = AgglomerativeClustering(n_clusters=10,
connectivity=connectivity,
linkage=linkage)
# Check that we obtain the same solution with early-stopping of the
# tree building
clustering.compute_full_tree = False
clustering.fit(X)
np.testing.assert_array_equal(clustering.labels_, labels)
clustering.connectivity = None
clustering.fit(X)
assert_true(np.size(np.unique(clustering.labels_)) == 10)
# Check that we raise a TypeError on dense matrices
clustering = AgglomerativeClustering(
n_clusters=10,
connectivity=sparse.lil_matrix(connectivity.toarray()[:10, :10]),
linkage=linkage)
assert_raises(ValueError, clustering.fit, X)
# Test that using ward with another metric than euclidean raises an
# exception
clustering = AgglomerativeClustering(n_clusters=10,
connectivity=connectivity.toarray(),
affinity="manhattan",
linkage="ward")
assert_raises(ValueError, clustering.fit, X)
# Test using another metric than euclidean works with linkage complete
for affinity in PAIRED_DISTANCES.keys():
# Compare our (structured) implementation to scipy
clustering = AgglomerativeClustering(n_clusters=10,
connectivity=np.ones(
(n_samples, n_samples)),
affinity=affinity,
linkage="complete")
clustering.fit(X)
clustering2 = AgglomerativeClustering(n_clusters=10,
connectivity=None,
affinity=affinity,
linkage="complete")
clustering2.fit(X)
assert_almost_equal(
normalized_mutual_info_score(clustering2.labels_,
clustering.labels_), 1)
def test_hdbscan_badargs():
assert_raises(ValueError, hdbscan, X='fail')
assert_raises(ValueError, hdbscan, X=None)
assert_raises(ValueError, hdbscan, X, min_cluster_size='fail')
assert_raises(ValueError, hdbscan, X, min_samples='fail')
assert_raises(ValueError, hdbscan, X, min_samples=-1)
assert_raises(ValueError, hdbscan, X, metric='imperial')
assert_raises(ValueError, hdbscan, X, metric=None)
assert_raises(ValueError, hdbscan, X, metric='minkowski', p=-1)
assert_raises(ValueError,
hdbscan,
X,
metric='minkowski',
p=-1,
algorithm='prims_kdtree')
assert_raises(ValueError,
hdbscan,
X,
metric='minkowski',
p=-1,
algorithm='prims_balltree')
assert_raises(ValueError,
hdbscan,
X,
metric='minkowski',
p=-1,
algorithm='boruvka_balltree')
assert_raises(ValueError,
hdbscan,
X,
metric='precomputed',
algorithm='boruvka_kdtree')
assert_raises(ValueError,
hdbscan,
X,
metric='precomputed',
algorithm='prims_kdtree')
assert_raises(ValueError,
hdbscan,
X,
metric='precomputed',
algorithm='prims_balltree')
assert_raises(ValueError,
hdbscan,
X,
metric='precomputed',
algorithm='boruvka_balltree')
assert_raises(ValueError, hdbscan, X, alpha=-1)
assert_raises(ValueError, hdbscan, X, alpha='fail')
assert_raises(Exception, hdbscan, X, algorithm='something_else')
assert_raises(TypeError, hdbscan, X, metric='minkowski', p=None)
assert_raises(ValueError, hdbscan, X, leaf_size=0)
def test_format_invariance_with_1d_vectors(name):
random_state = check_random_state(0)
y1 = random_state.randint(0, 2, size=(20, ))
y2 = random_state.randint(0, 2, size=(20, ))
y1_list = list(y1)
y2_list = list(y2)
y1_1d, y2_1d = np.array(y1), np.array(y2)
assert_array_equal(y1_1d.ndim, 1)
assert_array_equal(y2_1d.ndim, 1)
y1_column = np.reshape(y1_1d, (-1, 1))
y2_column = np.reshape(y2_1d, (-1, 1))
y1_row = np.reshape(y1_1d, (1, -1))
y2_row = np.reshape(y2_1d, (1, -1))
with ignore_warnings():
metric = ALL_METRICS[name]
measure = metric(y1, y2)
assert_allclose(metric(y1_list, y2_list),
measure,
err_msg="%s is not representation invariant with list"
"" % name)
assert_allclose(metric(y1_1d, y2_1d),
measure,
err_msg="%s is not representation invariant with "
"np-array-1d" % name)
assert_allclose(metric(y1_column, y2_column),
measure,
err_msg="%s is not representation invariant with "
"np-array-column" % name)
# Mix format support
assert_allclose(metric(y1_1d, y2_list),
measure,
err_msg="%s is not representation invariant with mix "
"np-array-1d and list" % name)
assert_allclose(metric(y1_list, y2_1d),
measure,
err_msg="%s is not representation invariant with mix "
"np-array-1d and list" % name)
assert_allclose(metric(y1_1d, y2_column),
measure,
err_msg="%s is not representation invariant with mix "
"np-array-1d and np-array-column" % name)
assert_allclose(metric(y1_column, y2_1d),
measure,
err_msg="%s is not representation invariant with mix "
"np-array-1d and np-array-column" % name)
assert_allclose(metric(y1_list, y2_column),
measure,
err_msg="%s is not representation invariant with mix "
"list and np-array-column" % name)
assert_allclose(metric(y1_column, y2_list),
measure,
err_msg="%s is not representation invariant with mix "
"list and np-array-column" % name)
# These mix representations aren't allowed
assert_raises(ValueError, metric, y1_1d, y2_row)
assert_raises(ValueError, metric, y1_row, y2_1d)
assert_raises(ValueError, metric, y1_list, y2_row)
assert_raises(ValueError, metric, y1_row, y2_list)
assert_raises(ValueError, metric, y1_column, y2_row)
assert_raises(ValueError, metric, y1_row, y2_column)
# NB: We do not test for y1_row, y2_row as these may be
# interpreted as multilabel or multioutput data.
if (name not in (MULTIOUTPUT_METRICS | THRESHOLDED_MULTILABEL_METRICS
| MULTILABELS_METRICS)):
assert_raises(ValueError, metric, y1_row, y2_row)
def test_multioutput_enetcv_error():
rng = np.random.RandomState(0)
X = rng.randn(10, 2)
y = rng.randn(10, 2)
clf = ElasticNetCV()
assert_raises(ValueError, clf.fit, X, y)
def test_validate_function():
"""Check that valid functions are accepted & invalid ones raise error"""
# Check arity tests
fun = make_function(function=_protected_sqrt, name='sqrt', arity=1)
# non-integer arity
assert_raises(ValueError, make_function, _protected_sqrt, 'sqrt', '1')
assert_raises(ValueError, make_function, _protected_sqrt, 'sqrt', 1.0)
# non-matching arity
assert_raises(ValueError, make_function, _protected_sqrt, 'sqrt', 2)
assert_raises(ValueError, make_function, maximum, 'max', 1)
# Check name test
assert_raises(ValueError, make_function, _protected_sqrt, 2, 1)
# Check return type tests
def bad_fun1(x1, x2):
return 'ni'
assert_raises(ValueError, make_function, bad_fun1, 'ni', 2)
# Check return shape tests
def bad_fun2(x1):
return np.ones((2, 1))
assert_raises(ValueError, make_function, bad_fun2, 'ni', 1)
# Check closure for negatives test
def _unprotected_sqrt(x1):
with np.errstate(divide='ignore', invalid='ignore'):
return np.sqrt(x1)
assert_raises(ValueError, make_function, _unprotected_sqrt, 'sqrt', 1)
# Check closure for zeros test
def _unprotected_div(x1, x2):
with np.errstate(divide='ignore', invalid='ignore'):
return np.divide(x1, x2)
assert_raises(ValueError, make_function, _unprotected_div, 'div', 2)
def test_multioutput_number_of_output_differ(name):
y_true = np.array([[1, 0, 0, 1], [0, 1, 1, 1], [1, 1, 0, 1]])
y_pred = np.array([[0, 0], [1, 0], [0, 0]])
metric = ALL_METRICS[name]
assert_raises(ValueError, metric, y_true, y_pred)
def test_unknown_strategey_regressor():
X = [[0]] * 5
y = [1, 2, 4, 6, 8]
est = DummyRegressor(strategy='gona')
assert_raises(ValueError, est.fit, X, y)
def check_sample_weight_invariance(name, metric, y1, y2):
rng = np.random.RandomState(0)
sample_weight = rng.randint(1, 10, size=len(y1))
# check that unit weights gives the same score as no weight
unweighted_score = metric(y1, y2, sample_weight=None)
assert_allclose(unweighted_score,
metric(y1, y2, sample_weight=np.ones(shape=len(y1))),
err_msg="For %s sample_weight=None is not equivalent to "
"sample_weight=ones" % name)
# check that the weighted and unweighted scores are unequal
weighted_score = metric(y1, y2, sample_weight=sample_weight)
# use context manager to supply custom error message
with assert_raises(AssertionError) as cm:
assert_allclose(unweighted_score, weighted_score)
cm.msg = ("Unweighted and weighted scores are unexpectedly almost "
"equal (%s) and (%s) for %s" %
(unweighted_score, weighted_score, name))
# check that sample_weight can be a list
weighted_score_list = metric(y1, y2, sample_weight=sample_weight.tolist())
assert_allclose(
weighted_score,
weighted_score_list,
err_msg=("Weighted scores for array and list "
"sample_weight input are not equal (%s != %s) for %s") %
(weighted_score, weighted_score_list, name))
# check that integer weights is the same as repeated samples
repeat_weighted_score = metric(np.repeat(y1, sample_weight, axis=0),
np.repeat(y2, sample_weight, axis=0),
sample_weight=None)
assert_allclose(weighted_score,
repeat_weighted_score,
err_msg="Weighting %s is not equal to repeating samples" %
name)
# check that ignoring a fraction of the samples is equivalent to setting
# the corresponding weights to zero
sample_weight_subset = sample_weight[1::2]
sample_weight_zeroed = np.copy(sample_weight)
sample_weight_zeroed[::2] = 0
y1_subset = y1[1::2]
y2_subset = y2[1::2]
weighted_score_subset = metric(y1_subset,
y2_subset,
sample_weight=sample_weight_subset)
weighted_score_zeroed = metric(y1, y2, sample_weight=sample_weight_zeroed)
assert_allclose(
weighted_score_subset,
weighted_score_zeroed,
err_msg=("Zeroing weights does not give the same result as "
"removing the corresponding samples (%s != %s) for %s" %
(weighted_score_zeroed, weighted_score_subset, name)))
if not name.startswith('unnormalized'):
# check that the score is invariant under scaling of the weights by a
# common factor
for scaling in [2, 0.3]:
assert_allclose(weighted_score,
metric(y1,
y2,
sample_weight=sample_weight * scaling),
err_msg="%s sample_weight is not invariant "
"under scaling" % name)
# Check that if number of samples in y_true and sample_weight are not
# equal, meaningful error is raised.
error_message = ("Found input variables with inconsistent numbers of "
"samples: [{}, {}, {}]".format(
_num_samples(y1), _num_samples(y2),
_num_samples(sample_weight) * 2))
assert_raise_message(ValueError,
error_message,
metric,
y1,
y2,
sample_weight=np.hstack(
[sample_weight, sample_weight]))
def test_quantile_strategy_empty_train():
est = DummyRegressor(strategy="quantile", quantile=0.4)
assert_raises(ValueError, est.fit, [], [])
def test_constants_not_specified_regressor():
X = [[0]] * 5
y = [1, 2, 4, 6, 8]
est = DummyRegressor(strategy='constant')
assert_raises(TypeError, est.fit, X, y)
