Refactor/1080 onehsot regressionmodels #1291
Conversation
…ot-regressionmodels
…reshaping depending on type of RegressionModel used (multi_model or one_shot) - files formating
Codecov ReportBase: 93.84% // Head: 93.87% // Increases project coverage by
Additional details and impacted files@@ Coverage Diff @@
## master #1291 +/- ##
==========================================
+ Coverage 93.84% 93.87% +0.02%
==========================================
Files 78 78
Lines 8503 8500 -3
==========================================
- Hits 7980 7979 -1
+ Misses 523 521 -2
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. ☔ View full report at Codecov. |
There was a problem hiding this comment.
This looks very solid to me as far as I can tell. Great job @eliane-maalouf 👏 and a super nice improvement!!
Before we can merge, could you add unit tests?
Also, you need to propagate the support to LinearRegressionModel, LGBM and Catboost (it's just about adding the parameters there).
- correct target column name for multi_models = False Co-authored-by: Julien Herzen <[email protected]>
…_models parameters to all RegressionModel models.
… output_chunk_len != 0)
…// updated rest of unittests for one_shot regression // formatting
Fixes #1080.
Summary
Refactors the RegressionModel class to provide a k_step ahead prediction with one model.
The option is still available to allow a model per future lag.
Other Information