search

unit8co / darts

A python library for user-friendly forecasting and anomaly detection on time series.
https://unit8co.github.io/darts/
Apache License 2.0
8.11k stars 884 forks source link

Support for Future Covariates & Probabilistic Forecasting in AutoCES #1712

Open arcgolatkar opened 1 year ago

arcgolatkar commented 1 year ago

Exogenous variables are supported for AutoCES in statsforecast library but the support for future covariates is missing in Darts. It would be great if you could add this feature.

dennisbader commented 1 year ago

Additionally, add probabilistic support for AutoCES as discussed here.

This requires raising the cap for statsforecast to >=1.5.0.

Beerstabr commented 1 year ago

I believe the same holds for AutoCES as did for AutoETS from Statsforecast. Although it is implied in the API reference of Statsforecast, AutoETS does not really support future covariates. You can input a X variable, however it will not be used as can be observed in the source code. I raised this issue here.

We could solve it the same way we solved it for AutoETS, namely by first regressing against the future covariates and then fitting AutoCES on the residuals.

dennisbader commented 1 year ago

@Beerstabr agreed, let's do it simarly as for AutoETS 👍