rsample::initial_split has the
strata param, but how would I block by >=1 variable?
I'm used to resampling from the mlr package, which includes stratification and blocking. In my case, I would like to stratify by the target variable, but also block by individual, given that there are multiple observations per individual. It's not clear how to block with
rsample::initial_split, based on the function docs or the tidymodels tutorial.
We'll be implementing this in the next week or so. It would be helpful if you could provide a small dummy data set that reflects your problem that we can use to test with. Do the stratification or blocking columns vary within a group/individual?
That's great! Sorry, but I can't provide a dataset from my own research. The iris dataset is used in the mlr docs: https://mlr.mlr-org.com/articles/tutorial/resample.html