Arbogodaï, a New Approach for Decision Trees

Decision tree methods generally suppose that the number of categories of the attribute to be predicted is fixed. Breiman et al., with their Twoing criterion in CART, considered gathering the categories of the predicted attribute into two superclasses. In this paper, we propose an extension of this method. We try to merge the categories in an optimal unspecified number of superclasses. Our method, called Arbogodai, allows during tree growing to group categories of the target variable as well as categories of the predictive attributes. At the end, the user can chose to generate either a set of single rules or or a set of multi-conclusion rules that provide interval like predictions.