get_fcast() is a two step process:
- time series cross validation: models are scored with WIS
top_n models seclected + ensembling for forecasting h weeks ahead.
For future implementation with models that may be slow, we want to seperate those two steps:
- becomes
cross_validation() (or ts_cv())
get_fcast() may require providing the models explicitely.
get_fcast()is a two step process:top_nmodels seclected + ensembling for forecastinghweeks ahead.For future implementation with models that may be slow, we want to seperate those two steps:
cross_validation()(orts_cv())get_fcast()may require providing the models explicitely.