Out-of-sample accuracy via modeltime.resample package

Machine Learning and Modeling
1

Hi,

In the context of model selection, I am performing cross-validation using the function modeltime_fit_resamples() available in the package modeltime.resample using the following code:

resamples_tscv <- time_series_cv(
    data        = df_train,
    date_var    = date, 
    assess      = "2 months",
    initial     = "6 years",
    skip        = "1 month",
    cumulative  = TRUE 
)

resamples_fitted <- models_tbl %>%
    modeltime_fit_resamples(
        resamples = resamples_tscv,
        control   = control_resamples(verbose = TRUE)
    )

resamples_fitted %>%
    modeltime_resample_accuracy() %>%
    table_modeltime_accuracy()

After fitting the best model on the whole training set, I would like to assess its accuracy for the 2-months ahead predictions on an independent test set. Basically, I am looking for a function similar to modeltime_fit_resamples() where the model is kept constant and not fitted on each resample. Does it exist?

Thanks in advance for your help,

Kind Regards,
Alessandro

2

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.