predict does not work with a crossvalidated tree?

Hey, I am trying to do the following: rtree = RegressionTree.fit(X,Y,'crossval','on'); pred = predict(rtree,X)

But unfortunately I get a error message.

Does anyone know what I can use instead of pred?

Thank you!

Answer 1

[Shashank Prasanna]

When you use the 'crossval' 'on' flag you get a partitioned model as an output. You can predict from this model using

kfoldpredict

http://www.mathworks.com/help/stats/classificationpartitionedmodel.kfoldpredict.html

[Tania]

Great, it worked!:) Do you know maybe also the other to this question: What is the difference between 'control depth' and 'pruning' in the 'improving classification and regression trees' documentation:

http://www.mathworks.de/de/help/stats/classification-trees-and-regression-trees.html#bsw6a96

I dont understand why they do both...?!?Both are trying to find the optimal depth through cross validation...

Thanks!