Customer classification in commercial bank based on rough set theory and fuzzy support vector machine

In the analysis of customer classification, redundant variables in the samples spoil the performance of the SVM classifier and reduce the recognition accuracy. On the other hand, we usually canpsilat label one customer as absolutely good who is sure to repay in time, or absolutely bad who will default certainly. In order to solve the problems mentioned above, this paper used rough sets (RS) as a preprocessor of SVM to select a subset of input variables and employ fuzzy support vector machine (FSVM), proposed in previous papers, to treat every sample as both positive and negative classes, but with different memberships. Additionally, the proposed RS-FSVM with membership based on affinity is tested on two different datasets. Then we compared the accuracies of proposed RS-FSVM model with other three models. Especially, in application of the proposed method, training sets are selected by increasing proportion. Experimental results showed that the RS-SVM model performed the best classification accuracy and generalization, implying that the hybrid of RS with fuzzy SVM model can serve as a promising alternative for customer classification.