Brandon Greenwell
Brandon Greenwell
Surrogate splits are always on by default, even when your training data are complete (you can turn them off via rpart.control if needed). The purpose here is so that you...
Extending to multiple predictors should be pretty straight forward. I'll try to work on this over the next couple of week and push an update as soon as I can....
I don't see that being an issue, but as a last resort, that is probably what I would do (but, of course, also citing the `propogate` package as the source...
Two issues in the above code: 1) `newdata` is not the second positional argument in **mlr**'s prediction method, so you'll need to name it; 2) you omitted the wrong column...
Sounds good @topepo, there's an option called `smaller_is_better` in the call to `vi_permute()` that defaults to `NULL` with some logic in the case where a user picks one of the...
Hey @topepo, sorry for the late reply. I'm not against making useful changes to the package as long as it doesn't break backward compatibility too much. And good call out...
Sounds good @topepo 👌
Just peeking through the package, it seems that using the `*_vec()` (e.g., `roc_auc_vec()`) family of functions would be relatively straightforward.
@topepo I'll be moving onto this package soon, just trying to tidy up some issues with [fastshap](https://bgreenwell.github.io/fastshap/) first.
Hey @topepo, started to integrate with **yardstick** if you want to take a look (changes are only on the devel branch and no unit tests or anything yet, but planning...