xpred.rpart malfunction

[aidacm]

Hi there!

I am trying to use the xpred.rpart function to generate an R2/AUC value for cross validation in order to stablish the best cp value but I have had some trouble with it.

In particular, in some cases, when the cp takes "small" values, the values generated by the function (i.e., the predicted values/class) do not make sense.

See the head of the generated matrix below:

As it can be seen, the first five columns contain values that make sense but, from the sixth on, they all contain zeros.

Is there a way to work this around?

Thank you for your help.