def test_ovr_multinomial_iris():
# Test that OvR and multinomial are correct using the iris dataset.
train, target = iris.data, iris.target
n_samples, n_features = train.shape
# The cv indices from stratified kfold (where stratification is done based
# on the fine-grained iris classes, i.e, before the classes 0 and 1 are
# conflated) is used for both clf and clf1
n_cv = 2
cv = StratifiedKFold(n_cv)
precomputed_folds = list(cv.split(train, target))
# Train clf on the original dataset where classes 0 and 1 are separated
clf = LogisticRegressionCV(cv=precomputed_folds)
clf.fit(train, target)
# Conflate classes 0 and 1 and train clf1 on this modified dataset
clf1 = LogisticRegressionCV(cv=precomputed_folds)
target_copy = target.copy()
target_copy[target_copy == 0] = 1
clf1.fit(train, target_copy)
# Ensure that what OvR learns for class2 is same regardless of whether
# classes 0 and 1 are separated or not
assert_array_almost_equal(clf.scores_[2], clf1.scores_[2])
assert_array_almost_equal(clf.intercept_[2:], clf1.intercept_)
assert_array_almost_equal(clf.coef_[2][np.newaxis, :], clf1.coef_)
# Test the shape of various attributes.
assert_equal(clf.coef_.shape, (3, n_features))
assert_array_equal(clf.classes_, [0, 1, 2])
coefs_paths = np.asarray(list(clf.coefs_paths_.values()))
assert_array_almost_equal(coefs_paths.shape, (3, n_cv, 10, n_features + 1))
assert_equal(clf.Cs_.shape, (10, ))
scores = np.asarray(list(clf.scores_.values()))
assert_equal(scores.shape, (3, n_cv, 10))
# Test that for the iris data multinomial gives a better accuracy than OvR
for solver in ['lbfgs', 'newton-cg', 'sag']:
max_iter = 100 if solver == 'sag' else 15
clf_multi = LogisticRegressionCV(solver=solver,
multi_class='multinomial',
max_iter=max_iter,
random_state=42,
tol=1e-2,
cv=2)
clf_multi.fit(train, target)
multi_score = clf_multi.score(train, target)
ovr_score = clf.score(train, target)
assert_greater(multi_score, ovr_score)
# Test attributes of LogisticRegressionCV
assert_equal(clf.coef_.shape, clf_multi.coef_.shape)
assert_array_equal(clf_multi.classes_, [0, 1, 2])
coefs_paths = np.asarray(list(clf_multi.coefs_paths_.values()))
assert_array_almost_equal(coefs_paths.shape,
(3, n_cv, 10, n_features + 1))
assert_equal(clf_multi.Cs_.shape, (10, ))
scores = np.asarray(list(clf_multi.scores_.values()))
assert_equal(scores.shape, (3, n_cv, 10))
def test_logistic_cv_score_does_not_warn_by_default():
lr = LogisticRegressionCV(cv=2)
lr.fit(X, Y1)
with pytest.warns(None) as record:
lr.score(X, lr.predict(X))
assert len(record) == 0
def test_ovr_multinomial_iris():
# Test that OvR and multinomial are correct using the iris dataset.
train, target = iris.data, iris.target
n_samples, n_features = train.shape
# The cv indices from stratified kfold (where stratification is done based
# on the fine-grained iris classes, i.e, before the classes 0 and 1 are
# conflated) is used for both clf and clf1
n_cv = 2
cv = StratifiedKFold(n_cv)
precomputed_folds = list(cv.split(train, target))
# Train clf on the original dataset where classes 0 and 1 are separated
clf = LogisticRegressionCV(cv=precomputed_folds)
clf.fit(train, target)
# Conflate classes 0 and 1 and train clf1 on this modified dataset
clf1 = LogisticRegressionCV(cv=precomputed_folds)
target_copy = target.copy()
target_copy[target_copy == 0] = 1
clf1.fit(train, target_copy)
# Ensure that what OvR learns for class2 is same regardless of whether
# classes 0 and 1 are separated or not
assert_array_almost_equal(clf.scores_[2], clf1.scores_[2])
assert_array_almost_equal(clf.intercept_[2:], clf1.intercept_)
assert_array_almost_equal(clf.coef_[2][np.newaxis, :], clf1.coef_)
# Test the shape of various attributes.
assert_equal(clf.coef_.shape, (3, n_features))
assert_array_equal(clf.classes_, [0, 1, 2])
coefs_paths = np.asarray(list(clf.coefs_paths_.values()))
assert_array_almost_equal(coefs_paths.shape, (3, n_cv, 10, n_features + 1))
assert_equal(clf.Cs_.shape, (10,))
scores = np.asarray(list(clf.scores_.values()))
assert_equal(scores.shape, (3, n_cv, 10))
# Test that for the iris data multinomial gives a better accuracy than OvR
for solver in ['lbfgs', 'newton-cg', 'sag', 'saga']:
max_iter = 2000 if solver in ['sag', 'saga'] else 15
clf_multi = LogisticRegressionCV(
solver=solver, multi_class='multinomial', max_iter=max_iter,
random_state=42, tol=1e-5 if solver in ['sag', 'saga'] else 1e-2,
cv=2)
clf_multi.fit(train, target)
multi_score = clf_multi.score(train, target)
ovr_score = clf.score(train, target)
assert_greater(multi_score, ovr_score)
# Test attributes of LogisticRegressionCV
assert_equal(clf.coef_.shape, clf_multi.coef_.shape)
assert_array_equal(clf_multi.classes_, [0, 1, 2])
coefs_paths = np.asarray(list(clf_multi.coefs_paths_.values()))
assert_array_almost_equal(coefs_paths.shape, (3, n_cv, 10,
n_features + 1))
assert_equal(clf_multi.Cs_.shape, (10,))
scores = np.asarray(list(clf_multi.scores_.values()))
assert_equal(scores.shape, (3, n_cv, 10))
def test_logistic_cv_mock_scorer():
class MockScorer(object):
def __init__(self):
self.calls = 0
self.scores = [0.1, 0.4, 0.8, 0.5]
def __call__(self, model, X, y, sample_weight=None):
score = self.scores[self.calls % len(self.scores)]
self.calls += 1
return score
mock_scorer = MockScorer()
Cs = [1, 2, 3, 4]
cv = 2
lr = LogisticRegressionCV(Cs=Cs, scoring=mock_scorer, cv=cv)
lr.fit(X, Y1)
# Cs[2] has the highest score (0.8) from MockScorer
assert lr.C_[0] == Cs[2]
# scorer called 8 times (cv*len(Cs))
assert mock_scorer.calls == cv * len(Cs)
# reset mock_scorer
mock_scorer.calls = 0
with pytest.warns(ChangedBehaviorWarning):
custom_score = lr.score(X, lr.predict(X))
assert custom_score == mock_scorer.scores[0]
assert mock_scorer.calls == 1
def test_ovr_multinomial_iris():
# Test that OvR and multinomial are correct using the iris dataset.
train, target = iris.data, iris.target
n_samples, n_features = train.shape
# Use pre-defined fold as folds generated for different y
cv = StratifiedKFold(target, 3)
clf = LogisticRegressionCV(cv=cv)
clf.fit(train, target)
clf1 = LogisticRegressionCV(cv=cv)
target_copy = target.copy()
target_copy[target_copy == 0] = 1
clf1.fit(train, target_copy)
assert_array_almost_equal(clf.scores_[2], clf1.scores_[2])
assert_array_almost_equal(clf.intercept_[2:], clf1.intercept_)
assert_array_almost_equal(clf.coef_[2][np.newaxis, :], clf1.coef_)
# Test the shape of various attributes.
assert_equal(clf.coef_.shape, (3, n_features))
assert_array_equal(clf.classes_, [0, 1, 2])
coefs_paths = np.asarray(list(clf.coefs_paths_.values()))
assert_array_almost_equal(coefs_paths.shape, (3, 3, 10, n_features + 1))
assert_equal(clf.Cs_.shape, (10, ))
scores = np.asarray(list(clf.scores_.values()))
assert_equal(scores.shape, (3, 3, 10))
# Test that for the iris data multinomial gives a better accuracy than OvR
for solver in ['lbfgs', 'newton-cg']:
clf_multi = LogisticRegressionCV(
solver=solver, multi_class='multinomial', max_iter=15
)
clf_multi.fit(train, target)
multi_score = clf_multi.score(train, target)
ovr_score = clf.score(train, target)
assert_greater(multi_score, ovr_score)
# Test attributes of LogisticRegressionCV
assert_equal(clf.coef_.shape, clf_multi.coef_.shape)
assert_array_equal(clf_multi.classes_, [0, 1, 2])
coefs_paths = np.asarray(list(clf_multi.coefs_paths_.values()))
assert_array_almost_equal(coefs_paths.shape, (3, 3, 10,
n_features + 1))
assert_equal(clf_multi.Cs_.shape, (10, ))
scores = np.asarray(list(clf_multi.scores_.values()))
assert_equal(scores.shape, (3, 3, 10))
def test_ovr_multinomial_iris():
# Test that OvR and multinomial are correct using the iris dataset.
train, target = iris.data, iris.target
n_samples, n_features = train.shape
# Use pre-defined fold as folds generated for different y
cv = StratifiedKFold(target, 3)
clf = LogisticRegressionCV(cv=cv)
clf.fit(train, target)
clf1 = LogisticRegressionCV(cv=cv)
target_copy = target.copy()
target_copy[target_copy == 0] = 1
clf1.fit(train, target_copy)
assert_array_almost_equal(clf.scores_[2], clf1.scores_[2])
assert_array_almost_equal(clf.intercept_[2:], clf1.intercept_)
assert_array_almost_equal(clf.coef_[2][np.newaxis, :], clf1.coef_)
# Test the shape of various attributes.
assert_equal(clf.coef_.shape, (3, n_features))
assert_array_equal(clf.classes_, [0, 1, 2])
coefs_paths = np.asarray(list(clf.coefs_paths_.values()))
assert_array_almost_equal(coefs_paths.shape, (3, 3, 10, n_features + 1))
assert_equal(clf.Cs_.shape, (10, ))
scores = np.asarray(list(clf.scores_.values()))
assert_equal(scores.shape, (3, 3, 10))
# Test that for the iris data multinomial gives a better accuracy than OvR
for solver in ['lbfgs', 'newton-cg']:
clf_multi = LogisticRegressionCV(solver=solver,
multi_class='multinomial',
max_iter=15)
clf_multi.fit(train, target)
multi_score = clf_multi.score(train, target)
ovr_score = clf.score(train, target)
assert_greater(multi_score, ovr_score)
# Test attributes of LogisticRegressionCV
assert_equal(clf.coef_.shape, clf_multi.coef_.shape)
assert_array_equal(clf_multi.classes_, [0, 1, 2])
coefs_paths = np.asarray(list(clf_multi.coefs_paths_.values()))
assert_array_almost_equal(coefs_paths.shape,
(3, 3, 10, n_features + 1))
assert_equal(clf_multi.Cs_.shape, (10, ))
scores = np.asarray(list(clf_multi.scores_.values()))
assert_equal(scores.shape, (3, 3, 10))
