Closed
Description
In kknn::train.kknn()
, the "optimal" value is found by optimizing the holdout set error. When we invoke that function, we don't have the holdout data, so we give it the training set (since we need to give it something).
It would be better to remove it than to keep it in and let tune_*()
choose it because it had the best resubstitution error.
Metadata
Metadata
Assignees
Labels
No labels