Imp/Add new model StatsForecastAutoTBATS

[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 #2609.

Summary

Add new model StatsForecastAutoTBATS from nixtla.

[codecov]

Codecov Report

Attention: Patch coverage is 88.88889% with 4 lines in your changes missing coverage. Please review.

Project coverage is 94.19%. Comparing base (441a58a) to head (93e209b).
Report is 1 commits behind head on master.

Files with missing lines	Patch %	Lines
darts/models/forecasting/sf_auto_tbats.py	91.17%	3 Missing ⚠️
darts/models/__init__.py	50.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2611      +/-   ##
==========================================
- Coverage   94.25%   94.19%   -0.07%     
==========================================
  Files         140      141       +1     
  Lines       15427    15463      +36     
==========================================
+ Hits        14541    14565      +24     
- Misses        886      898      +12     

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

[dennisbader]

Thanks a lot for this PR @cnhwl, it looks very good and is almost ready to be merged 🚀

Before merging could you address the points from below?

• add entry to CHANGELOG.md
• add model to the README.md
• add model to darts/userguide/covariates.md in this table
• run probabilistic tests by adding the model to darts/tests/models/forecasting/test_probabilistic_models.py

[cnhwl]

Thank you very much for your code review and guidance! @dennisbader. I'll finish the points above soon. 🤝
By the way, I submitted PR #2544 three months ago and am waiting for code review. Is there a timeline for this?

[cnhwl]

Hi! @dennisbader I'm trying to add the models to darts/tests/models/forecasting/test_probabilistic_models.py, but I'm finding that the random_state parameter doesn't exist for StatsForecastAutoTBATS, which is preventing it from passing the test_fit_predict_determinism. I would be very grateful if you could give me some help with this!

[dennisbader]

Thank you very much for your code review and guidance! @dennisbader. I'll finish the points above soon. 🤝 By the way, I submitted PR #2544 three months ago and am waiting for code review. Is there a timeline for this?

Hi @cnhwl, our capacity is currently a bit limited and we still need some time to get to. Sorry for that, but we will get there eventually!

[dennisbader]

Hi! @dennisbader I'm trying to add the models to darts/tests/models/forecasting/test_probabilistic_models.py, but I'm finding that the random_state parameter doesn't exist for StatsForecastAutoTBATS, which is preventing it from passing the test_fit_predict_determinism. I would be very grateful if you could give me some help with this!

I see, that is true :)

Actually, in my conformal prediction PR #2552, I introduce the @random_method that can be used by any model to control the randomness (see here). This will allow you to control the sampling.

Once we merge the PR (I think by end of next week), we can add this to our StatsForecastModels as well. Is it okay for you to wait until then?

Otherwise, the PR looks great and is ready to be merged :) Thanks a lot for the great work @cnhwl 🚀

[cnhwl]

Hi! @dennisbader I'm trying to add the models to darts/tests/models/forecasting/test_probabilistic_models.py, but I'm finding that the random_state parameter doesn't exist for StatsForecastAutoTBATS, which is preventing it from passing the test_fit_predict_determinism. I would be very grateful if you could give me some help with this!

I see, that is true :)

Actually, in my conformal prediction PR #2552, I introduce the @random_method that can be used by any model to control the randomness (see here). This will allow you to control the sampling.

Once we merge the PR (I think by end of next week), we can add this to our StatsForecastModels as well. Is it okay for you to wait until then?

Otherwise, the PR looks great and is ready to be merged :) Thanks a lot for the great work @cnhwl 🚀

Thanks again for your help! @dennisbader Of course! I am more than willing to wait for your PR #2552, so please let me know when that happens. 🤝

[dennisbader]

Hi @cnhwl, just wanted to let you know that PR #2552 (conformal prediction) has been merged now :)

[cnhwl]

Hi @dennisbader, I've resolved the code conflict and am ready to merge anytime now! 🚀

[dennisbader]

Looks great now, thanks @cnhwl 🚀

I can add probabilistic support with randomness handling later for all statsforecast models.