Can we add additional columns? #4
Description
To date, we have only modified the existing prediction columns. Users can circumvent this via the custom adjustment step.
It would be nice to be able to add applicability models to the postprocessor as well as conformal/bootstrap prediction interval objects.
We would need to modify the ptype within the container as well as maybe adding more options to the predict() method (such as the alpha level for prediction intervals).
It is more complex, but it may be worth it.
Just a note for future us: I considered adding methods for applicability scores. We'd need to update the `ptype` to include the columns of the training set predictors (the data would be an argument to such functions).
Originally posted by @topepo in #1 (comment)