Skip to content

Fixes _fit(X)#610

Merged
TonyBagnall merged 4 commits intomasterfrom
uea_dev
Jan 12, 2021
Merged

Fixes _fit(X)#610
TonyBagnall merged 4 commits intomasterfrom
uea_dev

Conversation

@TonyBagnall
Copy link
Copy Markdown
Contributor

Reference Issues/PRs

Fixes #608 . See also #596.

What does this implement/fix? Explain your changes.

scikit learn changed how KNN works in v0.24. _fit(X) was replaced by _fit(X, y=None), with a test for a _tag whether 'requires_y' is true. this is called in sklearn/neighbors/_base.py function_fit.

y is not actually required. Fixed by implementing
def _more_tags(self):
return {'requires_y': False}
this will overwrite the base class requires_y: True, since this function _get_tags is calls all _more_tags in the inheritance hierarchy in reverse order, from base class to self, using inspect.getmro, which seems to be pythons version of reflection.

Tony Bagnall added 4 commits January 11, 2021 21:21
scikit learn changed how KNN works in 0.24. _fit(X) was replaced by _fit(X, y=None), with a test for a _tage whether 'requires_y' is true. y is not actually required, so this should fix it.
scikit learn changed how KNN works in 0.24. _fit(X) was replaced by _fit(X, y=None), with a test for a _tage whether 'requires_y' is true. y is not actually required, so this should fix it.
scikit learn would fail these quality checks
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

[BUG] TypeError from KNeighborsTimeSeriesClassifier

1 participant