There are two ways to solve this problem:
1) CLASSREGTREE can be used to solve this classification problem because response values are ordinal. Convert the numerical YFIT output from CLASSREGTREE/EVAL to integer by calling ceil(YFIT)). Note that when many levels in the predictor variables, even when the number of levels, L, is less than 32, calling CLASSREGTREE with 'classification' method would be very slow, because it has to go through all 2^(L-1) possibilities.
2) NaiveBayes can be used as follows:
NB=NaiveBayes.fit([hobbies favFoods],ageGroups, 'dist','mvmn');
C= NB.predict(newData);
