[R] Rpart - best split selection for class method and Gin

Terry Therneau therneau at mayo.edu
Thu May 21 18:38:07 CEST 2009


 I would suggest looking at the documentation found in Technical Report #61 in 
our department series.
 
     http://mayoresearch.mayo.edu/mayo/research/biostat/
     
The formulas for priors, losses, etc in Gini splitting are such that I can't 
deduce them myself from the source code (and I wrote it), but need to have the
document.

	Terry T.




More information about the R-help mailing list