What if there are too many levels in a categorical predictor?
CART will only search over all possible subsets of a categorical predictor for a limited number of levels. Beyond a threshold set by computational feasibility, CART will simply reject the problem. You can control this limit with the BOPTION NCLASSES = m command, but be aware that for m larger than 15, computation times increase dramatically.
SOLUTION: Convert The Variable Into Dummies
The ideal solution is to work with a supercomputer implementation of Salford Systems CART, because this will provide the optimal tree. Other alternatives are compromises that might not yield satisfactory results. One such compromise is to break the categorical variable into a vector of dummies. For example, a 50-level occupation variable could be coded into 50 separate indicators.Steinberg, Dan and Phillip Colla. CART—Classification and Regression Trees. San Diego, CA: Salford Systems, 1997.