基于主成分分析和支持向量机的个人信用评估

Personal Credit Scoring Based on PCA and SVM

本文针对信用评估指标维数较高的问题,运用主成分分析与支持向量机理论建立了一个新的个人信用评估预测模型。为反映该模型在信用评估分类方面的优越性,又分别建立了基于神经网络、K近邻判别分析等多种理论的信用评估模型,并用同一组数据对不同的模型分别进行训练,然后比较其预测分类正确率。实验结果表明,基于主成分分析与支持向量机理论的个人信用评估模型具有较优的预测分类正确率。

This paper attempts to build up a new personal credit scoring model based on principal component analysis（PCA） and support vector machine（SVM）. In order to present the superiority of this model in consumer credit scoring,it also establishes several other personal credit sco- ring models based on these theories such as neural networks, K-neighbor discriminate analysis and so on, and compares the forecasting accuracy of＇these model through training by a same set of data. The experiment results show that the forecasting accuracy of the personal credit scoring model based on PCA-SVM is superior.