
I use the CHAID package to analyse my data: chaidcomplete < chaid(binaryvar ~ a+b+c+d, data = complete, control = ctrl)
The question is how I can use R to calculate precision, recall and F on the CHAID model and a test dataset?
I found ROCR package with prediction and performance functions, but haven't yet been able to figure it out with CHAID. Potentially I can apply chaidcomplete on test data and then extract the probability for each datapoint in testdata whether binaryvar realization is 1. But here the question is again how I can apply it on test data let alone extract the probabilities.
Many thanks for your help!
