基于主成分线性加权综合评价的信用评分方法及应用  

Credit Scoring via Principal Component Analysis Linear-weighted Comprehensive Assessment and Application

作　　者：李建平[1] 徐伟宣[1] 石勇[2] 

机构地区：[1]中国科学院科技政策与管理科学研究所,北京100080 [2]中国科学院研究生院,北京100039

出　　处：《系统工程》2004年第8期64-68,共5页Systems Engineering

基　　金：中国科学院院长基金资助项目(yjjz946);中科院科技政策与管理科学研究所所长基金资助项目(0343sz)

摘　　要：基于我国商业银行现有的信用卡评分标准和信用评分方法,提出一种基于主成分线性加权的综合评价的信用评分方法。其优点在于能够实现指标项权重的客观性、能方便地适应我国不同地区由于经济文化的差别而带来的信用环境不同以及一个地区由于人口漂移快而带来的评分变化。实证检验表明模型训练结果符合信用卡风险管理实际,测试结果显示有较好的应用前景,与现有银行评分标准对比的研究表明,本文的方法具有明显优势。According to the current scorecard of Chinese commercial bank, this paper presents a credit scoring method via (principal) component analysis linear-weighted comprehensive assessment(PCALWCA) to the Chinese commercial bank in (credit) card risk management. The merits of this method includes that it have an objective weights of original indexes and can easily adapt to the different economic & cultural environments in different regions and the population drift in same (region) which cause the disparity of credit scoring. The demonstration examination shows that the model fits with the (practice) of credit card risk management and the test result shows the model have an application foreground. The contrast (research) indicates that this method has better results than that of bank's current scoring method.

关 键 词：信用卡 风险管理 信用评分 主成分分析 

分 类 号：F830[经济管理—金融学]