def test_deprecated():
# Test whether the deprecated decorator issues appropriate warnings
# Copied almost verbatim from http://docs.python.org/library/warnings.html
# First a function...
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter("always")
@deprecated()
def ham():
return "spam"
spam = ham()
assert_equal(spam, "spam") # function must remain usable
assert_equal(len(w), 1)
assert_true(issubclass(w[0].category, DeprecationWarning))
assert_true("deprecated" in str(w[0].message).lower())
# ... then a class.
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter("always")
@deprecated("don't use this")
class Ham(object):
SPAM = 1
ham = Ham()
assert_true(hasattr(ham, "SPAM"))
assert_equal(len(w), 1)
assert_true(issubclass(w[0].category, DeprecationWarning))
assert_true("deprecated" in str(w[0].message).lower())
def test_randomized_svd_sign_flip_with_transpose():
# Check if the randomized_svd sign flipping is always done based on u
# irrespective of transpose.
# See https://github.com/scikit-learn/scikit-learn/issues/5608
# for more details.
def max_loading_is_positive(u, v):
"""
returns bool tuple indicating if the values maximising np.abs
are positive across all rows for u and across all columns for v.
"""
u_based = (np.abs(u).max(axis=0) == u.max(axis=0)).all()
v_based = (np.abs(v).max(axis=1) == v.max(axis=1)).all()
return u_based, v_based
mat = np.arange(10 * 8).reshape(10, -1)
# Without transpose
u_flipped, _, v_flipped = randomized_svd(mat, 3, flip_sign=True)
u_based, v_based = max_loading_is_positive(u_flipped, v_flipped)
assert_true(u_based)
assert_false(v_based)
# With transpose
u_flipped_with_transpose, _, v_flipped_with_transpose = randomized_svd(
mat, 3, flip_sign=True, transpose=True)
u_based, v_based = max_loading_is_positive(u_flipped_with_transpose,
v_flipped_with_transpose)
assert_true(u_based)
assert_false(v_based)
def test_uniform_weights():
# with uniform weights, results should be identical to stats.mode
rng = np.random.RandomState(0)
x = rng.randint(10, size=(10, 5))
weights = np.ones(x.shape)
for axis in (None, 0, 1):
mode, score = stats.mode(x, axis)
mode2, score2 = weighted_mode(x, weights, axis)
assert_true(np.all(mode == mode2))
assert_true(np.all(score == score2))
def test_resample():
# Border case not worth mentioning in doctests
assert_true(resample() is None)
# Check that invalid arguments yield ValueError
assert_raises(ValueError, resample, [0], [0, 1])
assert_raises(ValueError,
resample, [0, 1], [0, 1],
replace=False,
n_samples=3)
assert_raises(ValueError, resample, [0, 1], [0, 1], meaning_of_life=42)
# Issue:6581, n_samples can be more when replace is True (default).
assert_equal(len(resample([1, 2], n_samples=5)), 5)
def test_compute_class_weight():
# Test (and demo) compute_class_weight.
y = np.asarray([2, 2, 2, 3, 3, 4])
classes = np.unique(y)
cw = assert_warns(DeprecationWarning, compute_class_weight, "auto",
classes, y)
assert_almost_equal(cw.sum(), classes.shape)
assert_true(cw[0] < cw[1] < cw[2])
cw = compute_class_weight("balanced", classes, y)
# total effect of samples is preserved
class_counts = np.bincount(y)[2:]
assert_almost_equal(np.dot(cw, class_counts), y.shape[0])
assert_true(cw[0] < cw[1] < cw[2])
def test_make_rng():
# Check the check_random_state utility function behavior
assert_true(check_random_state(None) is np.random.mtrand._rand)
assert_true(check_random_state(np.random) is np.random.mtrand._rand)
rng_42 = np.random.RandomState(42)
assert_true(check_random_state(42).randint(100) == rng_42.randint(100))
rng_42 = np.random.RandomState(42)
assert_true(check_random_state(rng_42) is rng_42)
rng_42 = np.random.RandomState(42)
assert_true(check_random_state(43).randint(100) != rng_42.randint(100))
assert_raises(ValueError, check_random_state, "some invalid seed")
def test_max_feature_regression():
# Test to make sure random state is set properly.
X, y = datasets.make_hastie_10_2(n_samples=12000, random_state=1)
X_train, X_test = X[:2000], X[2000:]
y_train, y_test = y[:2000], y[2000:]
gbrt = GradientBoostingClassifier(n_estimators=100,
min_samples_split=5,
max_depth=2,
learning_rate=.1,
max_features=2,
random_state=1)
gbrt.fit(X_train, y_train)
deviance = gbrt.loss_(y_test, gbrt.decision_function(X_test))
assert_true(deviance < 0.5, "GB failed with deviance %.4f" % deviance)
def check_sample_int(sample_without_replacement):
# This test is heavily inspired from test_random.py of python-core.
#
# For the entire allowable range of 0 <= k <= N, validate that
# the sample is of the correct length and contains only unique items
n_population = 100
for n_samples in range(n_population + 1):
s = sample_without_replacement(n_population, n_samples)
assert_equal(len(s), n_samples)
unique = np.unique(s)
assert_equal(np.size(unique), n_samples)
assert_true(np.all(unique < n_population))
# test edge case n_population == n_samples == 0
assert_equal(np.size(sample_without_replacement(0, 0)), 0)
def test_clone():
# Tests that clone creates a correct deep copy.
# We create an estimator, make a copy of its original state
# (which, in this case, is the current state of the estimator),
# and check that the obtained copy is a correct deep copy.
from uplift.feature_selection import SelectFpr, f_classif
selector = SelectFpr(f_classif, alpha=0.1)
new_selector = clone(selector)
assert_true(selector is not new_selector)
assert_equal(selector.get_params(), new_selector.get_params())
selector = SelectFpr(f_classif, alpha=np.zeros((10, 2)))
new_selector = clone(selector)
assert_true(selector is not new_selector)
def test_probability_log():
# Predict probabilities.
clf = GradientBoostingClassifier(n_estimators=100, random_state=1)
assert_raises(ValueError, clf.predict_proba, T)
clf.fit(X, y)
assert_array_equal(clf.predict(T), true_result)
# check if probabilities are in [0, 1].
y_proba = clf.predict_proba(T)
assert_true(np.all(y_proba >= 0.0))
assert_true(np.all(y_proba <= 1.0))
# derive predictions from probabilities
y_pred = clf.classes_.take(y_proba.argmax(axis=1), axis=0)
assert_array_equal(y_pred, true_result)
def check_classification_toy(presort, loss):
# Check classification on a toy dataset.
clf = GradientBoostingClassifier(loss=loss,
n_estimators=10,
random_state=1,
presort=presort)
assert_raises(ValueError, clf.predict, T)
clf.fit(X, y)
assert_array_equal(clf.predict(T), true_result)
assert_equal(10, len(clf.estimators_))
deviance_decrease = (clf.train_score_[:-1] - clf.train_score_[1:])
assert_true(np.any(deviance_decrease >= 0.0))
leaves = clf.apply(X)
assert_equal(leaves.shape, (6, 10, 1))
def test_staged_functions_defensive():
# test that staged_functions make defensive copies
rng = np.random.RandomState(0)
X = rng.uniform(size=(10, 3))
y = (4 * X[:, 0]).astype(np.int) + 1 # don't predict zeros
for estimator in [
GradientBoostingRegressor(),
GradientBoostingClassifier()
]:
estimator.fit(X, y)
for func in ['predict', 'decision_function', 'predict_proba']:
staged_func = getattr(estimator, "staged_" + func, None)
if staged_func is None:
# regressor has no staged_predict_proba
continue
with warnings.catch_warnings(record=True):
staged_result = list(staged_func(X))
staged_result[1][:] = 0
assert_true(np.all(staged_result[0] != 0))
def check_warm_start_oob(name):
# Test that the warm start computes oob score when asked.
X, y = hastie_X, hastie_y
ForestEstimator = FOREST_ESTIMATORS[name]
# Use 15 estimators to avoid 'some inputs do not have OOB scores' warning.
clf = ForestEstimator(n_estimators=15,
max_depth=3,
warm_start=False,
random_state=1,
bootstrap=True,
oob_score=True)
clf.fit(X, y)
clf_2 = ForestEstimator(n_estimators=5,
max_depth=3,
warm_start=False,
random_state=1,
bootstrap=True,
oob_score=False)
clf_2.fit(X, y)
clf_2.set_params(warm_start=True, oob_score=True, n_estimators=15)
clf_2.fit(X, y)
assert_true(hasattr(clf_2, 'oob_score_'))
assert_equal(clf.oob_score_, clf_2.oob_score_)
# Test that oob_score is computed even if we don't need to train
# additional trees.
clf_3 = ForestEstimator(n_estimators=15,
max_depth=3,
warm_start=True,
random_state=1,
bootstrap=True,
oob_score=False)
clf_3.fit(X, y)
assert_true(not (hasattr(clf_3, 'oob_score_')))
clf_3.set_params(oob_score=True)
ignore_warnings(clf_3.fit)(X, y)
assert_equal(clf.oob_score_, clf_3.oob_score_)
def test_random_choice_csc(n_samples=10000, random_state=24):
# Explicit class probabilities
classes = [np.array([0, 1]), np.array([0, 1, 2])]
class_probabilites = [np.array([0.5, 0.5]), np.array([0.6, 0.1, 0.3])]
got = random_choice_csc(n_samples, classes, class_probabilites,
random_state)
assert_true(sp.issparse(got))
for k in range(len(classes)):
p = np.bincount(got.getcol(k).toarray().ravel()) / float(n_samples)
assert_array_almost_equal(class_probabilites[k], p, decimal=1)
# Implicit class probabilities
classes = [[0, 1], [1, 2]] # test for array-like support
class_probabilites = [np.array([0.5, 0.5]), np.array([0, 1/2, 1/2])]
got = random_choice_csc(n_samples=n_samples,
classes=classes,
random_state=random_state)
assert_true(sp.issparse(got))
for k in range(len(classes)):
p = np.bincount(got.getcol(k).toarray().ravel()) / float(n_samples)
assert_array_almost_equal(class_probabilites[k], p, decimal=1)
# Edge case probabilities 1.0 and 0.0
classes = [np.array([0, 1]), np.array([0, 1, 2])]
class_probabilites = [np.array([1.0, 0.0]), np.array([0.0, 1.0, 0.0])]
got = random_choice_csc(n_samples, classes, class_probabilites,
random_state)
assert_true(sp.issparse(got))
for k in range(len(classes)):
p = np.bincount(got.getcol(k).toarray().ravel(),
minlength=len(class_probabilites[k])) / n_samples
assert_array_almost_equal(class_probabilites[k], p, decimal=1)
# One class target data
classes = [[1], [0]] # test for array-like support
class_probabilites = [np.array([0.0, 1.0]), np.array([1.0])]
got = random_choice_csc(n_samples=n_samples,
classes=classes,
random_state=random_state)
assert_true(sp.issparse(got))
for k in range(len(classes)):
p = np.bincount(got.getcol(k).toarray().ravel()) / n_samples
assert_array_almost_equal(class_probabilites[k], p, decimal=1)
def test_probability_exponential():
# Predict probabilities.
clf = GradientBoostingClassifier(loss='exponential',
n_estimators=100,
random_state=1)
assert_raises(ValueError, clf.predict_proba, T)
clf.fit(X, y)
assert_array_equal(clf.predict(T), true_result)
# check if probabilities are in [0, 1].
y_proba = clf.predict_proba(T)
assert_true(np.all(y_proba >= 0.0))
assert_true(np.all(y_proba <= 1.0))
score = clf.decision_function(T).ravel()
assert_array_almost_equal(y_proba[:, 1], 1.0 / (1.0 + np.exp(-2 * score)))
# derive predictions from probabilities
y_pred = clf.classes_.take(y_proba.argmax(axis=1), axis=0)
assert_array_equal(y_pred, true_result)
def check_importances(name, criterion, X, y):
ForestEstimator = FOREST_ESTIMATORS[name]
est = ForestEstimator(n_estimators=20, criterion=criterion, random_state=0)
est.fit(X, y)
importances = est.feature_importances_
n_important = np.sum(importances > 0.1)
assert_equal(importances.shape[0], 10)
assert_equal(n_important, 3)
# XXX: Remove this test in 0.19 after transform support to estimators
# is removed.
X_new = assert_warns(DeprecationWarning,
est.transform,
X,
threshold="mean")
assert_less(0 < X_new.shape[1], X.shape[1])
# Check with parallel
importances = est.feature_importances_
est.set_params(n_jobs=2)
importances_parrallel = est.feature_importances_
assert_array_almost_equal(importances, importances_parrallel)
# Check with sample weights
sample_weight = check_random_state(0).randint(1, 10, len(X))
est = ForestEstimator(n_estimators=20, random_state=0, criterion=criterion)
est.fit(X, y, sample_weight=sample_weight)
importances = est.feature_importances_
assert_true(np.all(importances >= 0.0))
for scale in [0.5, 10, 100]:
est = ForestEstimator(n_estimators=20,
random_state=0,
criterion=criterion)
est.fit(X, y, sample_weight=scale * sample_weight)
importances_bis = est.feature_importances_
assert_less(np.abs(importances - importances_bis).mean(), 0.001)
def test_feature_importances():
X = np.array(boston.data, dtype=np.float32)
y = np.array(boston.target, dtype=np.float32)
for presort in True, False:
clf = GradientBoostingRegressor(n_estimators=100,
max_depth=5,
min_samples_split=2,
random_state=1,
presort=presort)
clf.fit(X, y)
assert_true(hasattr(clf, 'feature_importances_'))
# XXX: Remove this test in 0.19 after transform support to estimators
# is removed.
X_new = assert_warns(DeprecationWarning,
clf.transform,
X,
threshold="mean")
assert_less(X_new.shape[1], X.shape[1])
feature_mask = (clf.feature_importances_ >
clf.feature_importances_.mean())
assert_array_almost_equal(X_new, X[:, feature_mask])
def test_get_params():
test = T(K(), K())
assert_true('a__d' in test.get_params(deep=True))
assert_true('a__d' not in test.get_params(deep=False))
test.set_params(a__d=2)
assert_true(test.a.d == 2)
assert_raises(ValueError, test.set_params, a__a=2)
def test_clone_nan():
# Regression test for cloning estimators with default parameter as np.nan
clf = MyEstimator(empty=np.nan)
clf2 = clone(clf)
assert_true(clf.empty is clf2.empty)
def test_get_params_deprecated():
# deprecated attribute should not show up as params
est = DeprecatedAttributeEstimator(a=1)
assert_true('a' in est.get_params())
assert_true('a' in est.get_params(deep=True))
assert_true('a' in est.get_params(deep=False))
assert_true('b' not in est.get_params())
assert_true('b' not in est.get_params(deep=True))
assert_true('b' not in est.get_params(deep=False))