Skip to content

[WIP] FIX make sure sample_weight is taken into account by estimators #14246

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

Closed
wants to merge 4 commits into from
Closed

[WIP] FIX make sure sample_weight is taken into account by estimators #14246

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
6ab4fe6
FIX make sure sample_weight is taken into account by estimators
glemaitre Jul 3, 2019
a853011
typo
glemaitre Jul 3, 2019
1271d8d
improve readibility
glemaitre Jul 5, 2019
05ebcc6
Merge remote-tracking branch 'glemaitre/is/fix_sample_weight' into is…
glemaitre Jul 5, 2019
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: 38 additions & 0 deletions sklearn/utils/estimator_checks.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -78,6 +78,7 @@ def _yield_checks(name, estimator):
yield check_sample_weights_pandas_series
yield check_sample_weights_list
yield check_sample_weights_invariance
yield check_sample_weights_equivalence_sampling
yield check_estimators_fit_returns_self
yield partial(check_estimators_fit_returns_self, readonly_memmap=True)

Expand Down Expand Up @@ -631,6 +632,43 @@ def check_sample_weights_invariance(name, estimator_orig):
% name)


@ignore_warnings(category=(DeprecationWarning, FutureWarning))
def check_sample_weights_equivalence_sampling(name, estimator_orig):
# check that the estimators yield same results for
# over-sample dataset by indice filtering and using sample_weight
if (has_fit_parameter(estimator_orig, "sample_weight") and
not (hasattr(estimator_orig, "_pairwise")
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

why don't we use the pairwise tool to make it pairwise? Or do we not properly support sample weights then?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No idea, the base test was coming from the other sample_weight common test.

and estimator_orig._pairwise)):
# We skip pairwise because the data is not pairwise

estimator1 = clone(estimator_orig)
estimator2 = clone(estimator_orig)
set_random_state(estimator1, random_state=0)
set_random_state(estimator2, random_state=0)

if is_classifier(estimator1):
X, y = load_iris(return_X_y=True)
else:
X, y = load_boston(return_X_y=True)
y = enforce_estimator_tags_y(estimator1, y)

step = 2
indices = np.arange(start=0, stop=y.size, step=step)
sample_weight = np.zeros((y.size,))
sample_weight[::step] = 1.

estimator1.fit(X, y=y, sample_weight=sample_weight)
estimator2.fit(X[indices], y[indices])

err_msg = ("For {} does not yield to the same results when given "
"sample_weight and an up-sampled dataset")
for method in ["predict", "transform"]:
if hasattr(estimator_orig, method):
X_pred1 = getattr(estimator1, method)(X)
X_pred2 = getattr(estimator2, method)(X)
assert_allclose_dense_sparse(X_pred1, X_pred2, err_msg=err_msg)


@ignore_warnings(category=(DeprecationWarning, FutureWarning, UserWarning))
def check_dtype_object(name, estimator_orig):
# check that estimators treat dtype object as numeric if possible
Expand Down