Time Series cross-validation with exogenous covariates (TSLM) in fable package

Could you please help.
I am trying to do TS cross-validation using the template from Hyndman's course, 3rd edition, this section.

The pipeline is as follows:
Original tsibble > streched tsibble > model > forecast(h = 1) > accuracy()
The problem is when TSLM has exogenous covariates, then forecast() seems to need two arguments, h and "new data". Therefore, an error is generated.

I know it might be possible to get around it with tsCV() (used in Hyndman 2nd edition) but I was wondering if there is a more recent method that can take a stretched tsibble.



Referred here by Forecasting: Principles and Practice, by Rob J Hyndman and George Athanasopoulos

There's an example here: Rob J Hyndman - Time series cross-validation using fable

If you provide new_data, you don't need to provide h.

1 Like

Thanks a lot for replying.
I had a different question about your example, but I posted it on its page.

This topic was automatically closed 21 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.