The problem
I would like to set up a nested CV where the outer loop is 5-fold CV and the inner loop is a simple train-validate split. Specifically, I'm trying to do a nested CV where the splits are 60-20-20 train-validate-test, and it loops five times over the mutually exclusive test sets.
It seems the way to do this is to use nested_cv(mtcars, outside = vfold_cv(v = 5), inside = validation_split(prop = .75)); however, validation_split() has been deprecated.
Is there a better way to do this type of split with the new three-way split functions?
Thanks!
Reproducible example
nested_cv(mtcars, outside = vfold_cv(v = 5), inside = validation_split(prop = .75))
The problem
I would like to set up a nested CV where the outer loop is 5-fold CV and the inner loop is a simple train-validate split. Specifically, I'm trying to do a nested CV where the splits are 60-20-20 train-validate-test, and it loops five times over the mutually exclusive test sets.
It seems the way to do this is to use
nested_cv(mtcars, outside = vfold_cv(v = 5), inside = validation_split(prop = .75)); however,validation_split()has been deprecated.Is there a better way to do this type of split with the new three-way split functions?
Thanks!
Reproducible example