Credit Risk Assessment Using Rough Set Theory and GA-Based SVM

This paper applies a. classifier, hybridizing rough set approach and improved support vector machine(SVM) using genetic optimization algorithm (GA), to the study of credit risk assessment in commercial banks. We can get reduced information table, which implies that the number of evaluation criteria, such as financial ratios and qualitative variables is reduced with no information loss through rough set approach. And then, this reduced information table is used to develop classification rules and train SVM. Especially, in order to improve the assessment accuracy, GA is applied to optimize the parameters of SVM classifier. The rationale of our hybrid system is using rules developed by rough sets for an object that matches any of the rules and SVM for one that dose not match any of them. The effectiveness of our methodology was verified by experiments comparing traditional discriminant analysis (DA) model, BP neural networks (BPN) and standard SVM with our approach.