-
Notifications
You must be signed in to change notification settings - Fork 908
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.
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
Apply statsmodels-based ARIMA/VARIMA to new TS #1036
Conversation
… feat/apply-arima-to-new-ts
Codecov Report
@@ Coverage Diff @@
## master #1036 +/- ##
==========================================
- Coverage 93.75% 93.74% -0.01%
==========================================
Files 80 80
Lines 8160 8197 +37
==========================================
+ Hits 7650 7684 +34
- Misses 510 513 +3
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Cool addition to darts! 🚀 Left a few minor comments
Co-authored-by: Dustin Brunner <92083143+brunnedu@users.noreply.github.com>
Co-authored-by: Dustin Brunner <92083143+brunnedu@users.noreply.github.com>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good! Only a few minor comments and a suggestion name for the new class
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Looks good! I've identified a few opportunities for (small) performance improvements using copy=False
- there might be more that I missed!
Before merging, would you agree to check that we do not unnecessarily prevent historical forecasts with retrain=False
anymore? I think all it takes will likely be to override the method _supports_non_retrainable_historical_forecasts()
in TransferrableDualCovariatesForecastingModel
to return True
instead of False
.
@piaz97 the tests are failing because |
Co-authored-by: Julien Herzen <julien@unit8.co>
- Removed a print statement - Moved an error message that was misleading - Change type of staVARMA to ndarray, otherwise the result will be a multiindexed pandas df instead of the expected ndarray
I added the suggestions from the previous comments, added support for backtesting with |
Fixes #920 .
Summary
ARIMA and VARIMA models wrapping statsmodels implementations now support the prediction new TS (in addition to the one used during training).
Other Information
DualCovariatesForecastingModel
, and adding the support for aseries
parameter in thepredict()
method. Not satisfied with the new base class name,StatsmodelsDualCovariatesForecastingModel
, any better idea would be welcomed.ARIMA
andVARIMA
models.