Using Support Vector Machine in Fuzzy Association Rule Mining

Fuzzy rule based classification systems is one of the most popular in pattern classification problems. The rules in the fuzzy models can be weighted to show the importance of generated rules where all attributes in the antecedent part of the rules have been usually weighted equally. Whereas the contributed attributes in a fuzzy model may have different influences on the decision making, a new method based on support vector machine-recursive feature elimination (SVM-RFE) has been proposed in this study to show the effects of attributes by weighting factors. Apriori algorithm and fuzzy association rule mining (FARM) have been used to generate the suitable rules which are weighted by fuzzy support value. The combination of the proposed method for attribute weighting and fuzzy support value for weighting the generated rules have been used to discriminate the samples of two different well known datasets iris and wine. The results show that this simple method can increase the rate of accuracy and reduce the dependency of model to fuzzy support value in Apriori algorithm and the number of rules.