Hi!
I am trying to understand if there's a way of doing k-fold cross validation using the final model fitted with stacks. I'd like to compare the stacked model to the candidate ones, ideally using the same splits object that the candidate models were trained on.
Thanks for your help!
No. It currently uses the bootstrap to tune the ensemble model.
I'm not sure what the best approach here. I worry that using the same resamples that go into making the ensemble model would lead to significant bias.
The best thing to do would be leave out an additional slice of data and only use that to compare the models to the stacked ensemble.
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.