摘要
研究了个人信用风险的计量问题,构建了基于人工免疫机制的个人信用风险模型,提出了双边抗体人工免疫概率模型,利用商业银行实际数据进行了计算.应用ROC方法对模型的预测能力进行了检验,并与逻辑回归方法进行了比较,达到并超过了逻辑回归模型的预测水平.该模型系统不仅可以应用于个人信用的度量,也可以应用于公司类客户的信用风险的度量以及电信和公共服务等领域.
This thesis presents the approach to constructing consumer credit risk model by analysing personal credit risk for commercial banks.It also presents Bilateral Antibody Artificial Immune Probability Model and calculated the credit score using actual data from commercial bank.Test the forecast capability of model with ROC,and compare the result with that of logistic regression.This capability is sensitive to quantity of sample data,the more quantity are,the more forecast capability is.This model is well applied to forecast probability of default for consumer.This model can use other extensive domain.
出处
《系统工程理论与实践》
EI
CSCD
北大核心
2009年第12期88-93,共6页
Systems Engineering-Theory & Practice
基金
中财121人才工程青年博士发展基金(QBG0701)
国家自然科学基金(70773124)
中央财经大学"211工程"三期
关键词
双边抗体
人工免疫
概率模型
违约概率
bilateral antibody
artificial immunity
probability model
probability of default
