SAFARI: A structured approach for automatic rule induction

This paper describes a new algorithm for obtaining rules automatically from training examples. The algorithm is applicable to examples involving both objects with discrete and continuous-valued attributes. The paper explains a new quantization procedure for continuous-valued attributes and shows how appropriate ranges of values of various attributes are obtained. The algor ithm uses a decision-tree-based approach for obtaining rules, but unlike ot her tree-based algorithms such as ID3, it allows more than one attribute at a node which greatly improves its performance. The ability of the algorith m to obtain a measure of partial match further enhances its generalization characteristic. The algorithm produces the same rules irrespective of the o rder of presentation of training examples. The algorithm has been demonstra ted on classification problems. The results have compared favorably with th ose obtained by existing inductive learning algorithms.