摘要
对原蚁群算法的转移概率和信息素更新机制进行改进,并首次将蚁群算法应用于商业银行的信用风险评估问题,取得满意的结果。通过将计算结果与回归分类算法、判别分析和遗传规则进行比较,表明应用该算法解决商业信用风险问题更加有效。
Ants algorithm was applied to credit risk assessment in commercial banks and satisfactory results were obtained. The original ants algorithm are improved in two aspects: transition probability and pheromone trail update mechanism. Comparison of results with those of recursive partition algorithm, discriminate analysis, and genetic programming suggest that the modified model is more effective.
出处
《天津大学学报(社会科学版)》
2005年第2期81-85,共5页
Journal of Tianjin University:Social Sciences
基金
国家杰出青年科学基金资助项目(70225002)
国家自然科学基金资助项目 (70041039 )
教育部跨世纪优秀人才基金资助项目(9901)
教育部优秀青年教师教学科研奖励基金资助项目(001-28).
关键词
蚁群算法
信用风险评估
商业银行
数据
ants algorithm
credit risk assessment
commercial banks
data