Dear R-List,
Is there any way to handle NA's within tree objects themselves in
terms of the predict.tree( )? I understand that if there is an NA at a
split in a certain variable, the prediction is classified at that node and
not dropped further down the tree. Are there any other and possibly more
robust ways to handle NAs within tree objects? Note: I only would like
information on tree( ) objects,random forest imputation et cetera are not
wanted in this application.
Thank you in advance,
Ken
[[alternative HTML version deleted]]
