近似支持向量机对个人住房贷款信用风险预测

为了降低银行的放贷风险，在构建商业银行个人住房贷款信用风险评价指标基础上，采用机器学习原理中的近似支持向量机（Proximal Support Vector Machines，PSVM）模型对某商业银行西安市场的个人住房贷款借款人数据进行实证分析，研究中个人住房贷款借款人的各项指标作为属性矩阵，借款人是否违约作为判别矩阵，利用260个样本的训练集获得最优超平面，再对40个样本的测试集进行预测，结果表明PSVM模型在预测商业银行个人住房贷款信用风险时的正确率达到了87．5％．

PSVM Model Based Forecasting of the Credit Risk of Individual Housing Loan

Based on the establishment of the index system of credit risk of individual housing loans in commercial banks,on empirical analysis was made of the data of borrowers of housing loans in a certain commercial bank in Xi＇ an by using proximal support vector machine （PSVM） in machine learning. In the study, the borrowers＇s indexes were set as the attribute matrix and whether the borrowers defaultor not was set as the discriminant matrix. 260 samples in the training set were used to get an optimal hyperplane,and then a forecast was made to the 40 samples in the testing set. The result suggests that the PSVM model achieves higher accuracy in the forecast of the credit risk of individual housing loans.