A Support Vector Machine Based Method For Credit Risk Assessment
The credit card industry has been growing rapidly in recent years, and credit risk assessment becomes critically important for financial companies. In this paper, a novel support vector machine (SVM) based ensemble model is proposed for credit risk assessment. In the proposed method, principles component analysis (PCA) is firstly employed for credit feature selection. Secondly, SVMs with different kernels are trained by using genetic algorithm (GA) to optimize the parameters, and the corresponding assessment results are obtained. Thirdly, all results produced by different SVMs are combined by several ensemble strategies. Finally, an optimal ensemble strategy is selected for credit scoring. For validation, two real world credit datasets are used to test the effectiveness and efficiency of our proposed method. The experiment results find that our proposed ensemble model outperforms commonly used credit scoring tools. The findings of the study reveal the support vector machine based ensemble method to be a promising alternative for credit scoring.
support vector machine ensemble method credit risk assessment credit scoring
Wei Xu1 Shenghu Zhou Dongmei Duan Yanhui Chen
Key Laboratory of Data Engineering and Knowledge Engineering (Renmin University of China), Ministry School of Information, Renmin University of China, Beijing 100872, China
国际会议
2010 IEEE International Conference on e-Business Engineering(2010年电子商务工程国际研讨会 ICEBE 2010)
上海
英文
50-55
2010-11-10（万方平台首次上网日期，不代表论文的发表时间）