摘要
分类树是一种计算机实现、基于统计理论的非参数的识别技术 .该技术可应用于解决贷款 5分类实施中从业人员分析判断能力欠缺的问题 .实证分析表明该方法比线性判别分析方法的准确率高 ;针对进一步提高分类准确率 。
Classification tree is a promising approach in credit risk analysis as a statistic based pattern recognition computer technique. In the paper, we try to apply it to lower discrimnant ability of bank personnel in the process of five loan classification. Empirical test shows that classification tree outperforms the discrimnant analysis method. At last, the paper analyses the effect to accuracy of primary sample set and variable structure.
出处
《系统工程学报》
CSCD
2001年第4期282-288,共7页
Journal of Systems Engineering
基金
国家自然科学基金资助项目 (79713 0 0 7)
关键词
商业银行
贷款
分类树
统计理论
信贷
five loan classification
classification tree
discrimnant analysis