基于SVM的个人房贷信用评估数据分析

Individual Housing Loan Credit Scoring Data Analysis Research Based On SVM

采用SVM的序列最小最优化算法(SMO)作为训练算法对商业银行个人房贷信用评估数据进行分析,着重探讨了在个人房贷信用评估中分别应用径向基核函数参数和SMO训练算法中的参数调整对准确度的影响;通过银行实际数据集将该算法与C4.5和神经网络进行了比较,支持向量机对个人信用评估的总精度高于其他两种算法;支持向量机对实际的住房抵押贷款数据进行信用评估效果较好,且参数调整对试验结果有影响。

This paper selects radialbasis funetion （RBF） kernel function in SMO for personal house loan credit evaluation in business bank, emphasizes on discussing the effects of RBF and SMO to accuracy. In comparison with SVM, C4.5 and nearal network. SVM has higher evaluation accuraly than the other two methods. SVM had a better credit evaluation effevt to the house mortgage loan, parameter variation will affect the experimental result as well.