Enhancing Fuzzy Rule Extraction Based on Rough Set Theory and Entropy

Rule extraction is an important theme in data mining. Fuzzy set theory(FST) and rough set theory(RST) are two common technologies frequently applied to data mining tasks. Decision induction is one of common approaches for extracting rules in data mining. Integrating the advantages of FST and RST, this paper proposes a hybrid system to efficiently extract decision rules from a decision table. Through fuzzy sets, numeric attributes can be represented by fuzzy numbers, interval values as well as crisp values. Second, the paper proposes to utilize information gain for distinguishing importance among attributes. Then, by applying rough set approach, a decision table can be reduced by removing redundant attributes without any information loss. Finally, decision rules can be extracted from the equivalence classes. An experiment result is also presented to show the applicability of the proposed method.