【摘要】 以武汉地区部分高校2782个国家助学贷款的还贷及违约数据为样本,首先应用因子分析法,在每个样本的12个属性中,提取7个主因子,再将实际数据的70%作为训练样本、30%作为测试样本,分别用Fisher判别法以及加权核Fisher判别法进行判别分类。结果表明,Fisher判别法的正判率为54.08%,而加权核Fisher判别法正判率高达74%。该项研究对银行贷款的决策和管理具有重要指导意义,其原理方法也可应用于其他类似问题的研究。更多还原

【Abstract】 This paper takes 2782 data of non-defaulted and defaulted state-subsidized student loan in a university as samples.Firstly,by using Factor Analysis,7 factors are picked up from original 12 attributes of each sample.Then 70% data are served as training samples and 30% data are served as test samples.Furthermore,Fisher Discriminant and Weighted Kernel Fisher Discriminant are respectively used to classify these data.The result indicates that the accuracy rate of Fisher Discriminant is 54.08%,while the accuracy rate of Weighted Kernel Fisher Discriminant reaches 74%.This research has guiding significance for banks in making decision on the loan as well as managing the loan,and the principle and method can also be applied into the others like this problem.更多还原