Hi,
I am trying to understand how the fit_members function in the stacks package works. After passing the candidate models and fitting them to the training set, how can we then fit the ensemble to the entire data ?
Thanks for your help!
That's what fit_members() does. The model objects that go into the data stack used cross-validation to estimate the models. Using those results, the stacking model chooses which of the members to keep in the ensemble. After it chooses the models to use, fit_member() generates their model fits on the original training set.