fix the sample_weights bug

[cnhwl]

Checklist before merging this PR:

• Mentioned all issues that this PR fixes or addresses.
• Summarized the updates of this PR under Summary.
• Added an entry under Unreleased in the Changelog.

Fixes #2439.

Summary

fixes the sample_weights bug by making conditional judgement on whether sample_weights is none and whether the parameter sample_weights exists in the model fit method.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 93.80%. Comparing base (e8b5daa) to head (44aa8a2).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2445      +/-   ##
==========================================
- Coverage   93.80%   93.80%   -0.01%     
==========================================
  Files         139      139              
  Lines       14680    14678       -2     
==========================================
- Hits        13771    13768       -3     
- Misses        909      910       +1     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[dennisbader]

Thanks for the fix and the contribution @cnhwl 🚀 It's highly appreciated :)

I just had a minor suggestion for code. And there are some things that we should address before merging:

• add a unit test to darts/tests/models/forecasting/test_regression_models.py that checks that it works as expected with models that don't support sample weights
• in darts/utils/multioutput.py - MultiOutputRegressor.fit() we currently raise an error if the user provided sample_weight but the model doesn't support it. The MultiOutputRegressor is applied if we need to predict more than one target component and / or output_chunk_length > 1. Ideally, we should have the same behavior there as well. So if you can replace raising an error with a warning, that should be fine. (these lines here)
• can you add an etry to our CHANGELOG.md under the Unreleased / Fixed section?

[dennisbader]

Thanks for the updates @cnhwl 🚀

I took the liberty to apply the last open points directly (e.g., adaptions to MultiOutputRegressor, and adding the KNeighborsRegressor only to the sample weights tests).

Looks great now! We can merge after all tests have passed 💯

[cnhwl]

Thank you so much! @dennisbader ❤️

I have benefited a lot from your code. It's been a pleasure working with you on this merge.

Looking forward to contributing more to the darts library! 🚀