基于LS-SVM的个人信用评估

个人信用评估一直是金融业研究的重要内容,已取得一定的研究成果,但与现实的业务发展要求还有一段距离。本文在信用评估中应用了一种新的方法———最小二乘支持向量机方法(LS-SVM),该方法具有专门针对小样本、算法复杂度与样本维数无关、处理非线性等优点。使用真实的信用卡数据实证结果表明,本方法具有较好的预测能力,在与Logistic方法进行个人信用评估的对比研究中,该方法具有明显的优势。

Personal credit assessment based on LS-SVM

Personal credit assessment has attached lots of researchers in the financial industry.At present,they have achieved some better results,but we have lots of work to do because of the development of financial transaction.In this paper,we apply LS-SVM(Least Square Support Vector Machine) to the personal credit assessment.This algorithm has many advantages,such as dealing with a finite sample,nonlinear problems,and the complexity of algorithm is irrelevant to the dimension of sample.Experiments based on the real data set demonstrate the LS-SVM method has outstanding assessment ability.Compared with the class Logistic method,it also shows the predominance of the LS-SVM method in predicting accuracy.