Hi! Does anyone know of a vignette or example that implements loo_cv into a tidymodel workflow for an ML method? Ineed to quantify my prediction error for each record (my dataset is about the same size as mtcars) so vfold_cv isn't quite what I'm looking for.
The examples using vfold_cv are great, but if I try to use an loo_cv object with
fit_resamples I get the error that loocv is not currently supported by tune. I see that the former of these creates data "folds" while the other creates "Resamples". If I try to create a recipe and bake it with the splits from loocv I get an error that "x should be an rsplit" object.
I can get loo_cv to work for the case of a traditional linear regression if I embed my whole process into a map() function. However, I can't quite wrap my head around doing this for an ML model. And it seems rather clunky and hacky.
Apologies that I do not have a reproducible example, but this question is more broad and I'm not sure how to do that.
I have done lots of searching online (and posted to SO) but have found no good resources for implementing the loo_cv into tidymodels flow with an ML model.