Skip to content

FIX faulty test cross_validate that used Lasso while intending to use SVC #25456

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Jan 24, 2023
Merged

FIX faulty test cross_validate that used Lasso while intending to use SVC #25456

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
8212a77
Fixed faulty test, by applying cross-validation to SVC for classifica…
shadyelgewily-slimstock Jan 23, 2023
43ca895
Merge branch 'main' into resolve_faulty_test_#25444
shadyelgewily-slimstock Jan 23, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
38 changes: 26 additions & 12 deletions sklearn/model_selection/tests/test_validation.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -426,8 +426,9 @@ def test_cross_validate():
train_r2_scores = []
test_r2_scores = []
fitted_estimators = []

for train, test in cv.split(X, y):
est = clone(reg).fit(X[train], y[train])
est = clone(est).fit(X[train], y[train])
train_mse_scores.append(mse_scorer(est, X[train], y[train]))
train_r2_scores.append(r2_scorer(est, X[train], y[train]))
test_mse_scores.append(mse_scorer(est, X[test], y[test]))
Expand All @@ -448,11 +449,14 @@ def test_cross_validate():
fitted_estimators,
)

check_cross_validate_single_metric(est, X, y, scores)
check_cross_validate_multi_metric(est, X, y, scores)
# To ensure that the test does not suffer from
# large statistical fluctuations due to slicing small datasets,
# we pass the cross-validation instance
check_cross_validate_single_metric(est, X, y, scores, cv)
check_cross_validate_multi_metric(est, X, y, scores, cv)


def check_cross_validate_single_metric(clf, X, y, scores):
def check_cross_validate_single_metric(clf, X, y, scores, cv):
(
train_mse_scores,
test_mse_scores,
Expand All @@ -465,12 +469,22 @@ def check_cross_validate_single_metric(clf, X, y, scores):
# Single metric passed as a string
if return_train_score:
mse_scores_dict = cross_validate(
clf, X, y, scoring="neg_mean_squared_error", return_train_score=True
clf,
X,
y,
scoring="neg_mean_squared_error",
return_train_score=True,
cv=cv,
)
assert_array_almost_equal(mse_scores_dict["train_score"], train_mse_scores)
else:
mse_scores_dict = cross_validate(
clf, X, y, scoring="neg_mean_squared_error", return_train_score=False
clf,
X,
y,
scoring="neg_mean_squared_error",
return_train_score=False,
cv=cv,
)
assert isinstance(mse_scores_dict, dict)
assert len(mse_scores_dict) == dict_len
Expand All @@ -480,27 +494,27 @@ def check_cross_validate_single_metric(clf, X, y, scores):
if return_train_score:
# It must be True by default - deprecated
r2_scores_dict = cross_validate(
clf, X, y, scoring=["r2"], return_train_score=True
clf, X, y, scoring=["r2"], return_train_score=True, cv=cv
)
assert_array_almost_equal(r2_scores_dict["train_r2"], train_r2_scores, True)
else:
r2_scores_dict = cross_validate(
clf, X, y, scoring=["r2"], return_train_score=False
clf, X, y, scoring=["r2"], return_train_score=False, cv=cv
)
assert isinstance(r2_scores_dict, dict)
assert len(r2_scores_dict) == dict_len
assert_array_almost_equal(r2_scores_dict["test_r2"], test_r2_scores)

# Test return_estimator option
mse_scores_dict = cross_validate(
clf, X, y, scoring="neg_mean_squared_error", return_estimator=True
clf, X, y, scoring="neg_mean_squared_error", return_estimator=True, cv=cv
)
for k, est in enumerate(mse_scores_dict["estimator"]):
assert_almost_equal(est.coef_, fitted_estimators[k].coef_)
assert_almost_equal(est.intercept_, fitted_estimators[k].intercept_)


def check_cross_validate_multi_metric(clf, X, y, scores):
def check_cross_validate_multi_metric(clf, X, y, scores, cv):
# Test multimetric evaluation when scoring is a list / dict
(
train_mse_scores,
Expand Down Expand Up @@ -541,15 +555,15 @@ def custom_scorer(clf, X, y):
if return_train_score:
# return_train_score must be True by default - deprecated
cv_results = cross_validate(
clf, X, y, scoring=scoring, return_train_score=True
clf, X, y, scoring=scoring, return_train_score=True, cv=cv
)
assert_array_almost_equal(cv_results["train_r2"], train_r2_scores)
assert_array_almost_equal(
cv_results["train_neg_mean_squared_error"], train_mse_scores
)
else:
cv_results = cross_validate(
clf, X, y, scoring=scoring, return_train_score=False
clf, X, y, scoring=scoring, return_train_score=False, cv=cv
)
assert isinstance(cv_results, dict)
assert set(cv_results.keys()) == (
Expand Down