Add support for static covariates and time-varying covariates that are known into the future for time series tasks. This corresponds to static_covariates and future_covariates in the terminology used by Darts.
Update AutoGluon code to handle static & time-varying covariates.
Fix a potential bug where the id_column in a time series dataset could be parsed as a str or an int in different parts of the code, leading to incorrect calculation of the seasonal error.
To do:
[ ] Explain how to create datasets with covariates in the documentation.
This PR introduces following changes:
static_covariatesandfuture_covariatesin the terminology used by Darts.id_columnin a time series dataset could be parsed as astror anintin different parts of the code, leading to incorrect calculation of the seasonal error.To do: