rpart - how to define cp values to be cross-validated over

Question

I am developing a regression model using the rpart function in R. This function cross-validates over the parameter cp.

Is there a way to control what values of cp and how many are cross-validated over?

Answer 1

You can do something like

fit <- rpart(Kyphosis ~ Age + Number + Start, data = kyphosis,
             control = rpart.control(cp = 0.05))

Also take a look at the rpart.control options http://www.inside-r.org/r-doc/rpart/rpart.control

It would probably be much easier to use caret

library(caret)
data(iris)
tc <- trainControl("cv",10)
rpart.grid <- expand.grid(.cp=0.2) 
train.rpart <- train(Kyphosis ~ Age + Number + Start,
                     data=kyphosis, method="rpart",
                     trControl=tc,  tuneGrid=rpart.grid)