Data Science Desktop Survival Guide
by Graham Williams
Complexity Parameter 0
We can set the cp= argument to be 0, so that no pruning of the tree is performed.
tmodel <- rpart(form, ds[vars], control=rpart.control(cp=0))
Notice that as we continue to build the model, by recursive partitioning, the model gets more complex but the performance does not improve, and in fact over time the model performance starts to deteriorate because of overfitting.