[pymvpa] TreeClassifier and complicated decision trees

[Thorsten Kranz]

Hi all,

I have a question regarding the TreeClassifier. I'm still using good
old 0.4-branch.

I try to reproduce a rather complicated decision tree with a
TreeClassifier. If I have e.g. four labels, my tree might look like:

         /   \
        3   / \
           1  /\
             2  4

(Hope my ASCII-drawing works...)

I have two questions now:
1) How do I define the corresponding classifier for this?
   TreeClassifier(SVM(),
{"g1":([3],SVM()),"g2":(