Fuzzy modeling by induction and pruning of decision trees

摘要

Presents a Fuzzy ID3 algorithm that generates a fuzzy rule base from a set of input-output data. The rules are induced by the ID3 algorithm of Quinlan (1986). Pruning conditions are presented that assist in the elimination of irrelevant attributes and simplification of the rule base. The algorithm is applied to the modeling of an ARMA process with both known and unknown model order.