基于支持向量机的个人信用评估模型及最优参数选择研究  

A Study of Personal Credit Scoring Models on Support Vector Machine with Optimal Choice of Kernel Function Parameters

作　　者：肖文兵[1] 费奇[1] 

机构地区：[1]华中科技大学系统工程研究所,武汉430074

出　　处：《系统工程理论与实践》2006年第10期73-79,共7页Systems Engineering-Theory & Practice

基　　金：国家自然科学基金(70171015)

摘　　要：运用基于支持向量机理论试图建立一个新的个人信用评估预测方法,以期取得更好的预测分类能力.为了达到这个目标及保证可靠性,研究中使用网格5-折交叉确认来寻找不同核函数的最优参数.为了进一步评价SVM分类准确性,我们在本文最后对SVM方法与线性判别分析,Logistic回归分析,最近邻,分类回归树及神经网络进行了比较,结果表明,SVM有很好的预测效果.As credit industry has expanded rapidly over last several years, credit scoring models have drawn a lot of research interests in previous literature. Recent studies have shown that machine learning techniques achieved better performance than traditional statistical ones. This paper applies support vector machines （SVMs） to the credit scoring prediction problem in an attempt to suggest a new model with better classification accuracy. To serve this purpose, we use a grid search technique using 5-fold cross-validation to fred out the optimal parameter values of various kernel function of SVM. In addition, to evaluate the prediction accuracy of SVM, we compare its performance with those of linear discfiminant analysis （LDA）, logistic regression analysis （Logit）, K-nearest neighbours （K-NN）, classification and regression tree and neural networks （ANN）. The experiment results show that SVM have a very good prediction accuracy.

关 键 词：信用评估 支持向量机(SVM) 神经网络(NN) 5-折交叉确认 

分 类 号：TP183[自动化与计算机技术—控制理论与控制工程]