Crossvalidation of Classification Trees?

Hi there, I want to perform a crossvalidation of a decision tree built with the CART algorithm, i.e. I want to randomly take out 20% (or 10% which is better?) from my dataset for the evaluation and thus build the tree with the residual 80% (or 90%).Is there a function in matlab that does this for me? I found "crossval" but I am not sure how the classification is done (is it also done according to CART?) Also, what does 10 fold cross-validation mean? Do I have to manually create a training and a test-data set if I want to use crossval?

Answer 1

[Richard Willey]

I recommend that you look at the following example from the file exchange

http://www.mathworks.com/matlabcentral/fileexchange/28770-introduction-to-classification