主管:中国科学院
主办:中国优选法统筹法与经济数学研究会
   中国科学院科技战略咨询研究院

中国管理科学 ›› 2015, Vol. 23 ›› Issue (4): 39-45.doi: 10.16381/j.cnki.issn1003-207x.2015.04.005

• 论文 • 上一篇    下一篇

基于差分进化自动聚类的信用风险评价模型研究

张大斌1,2, 周志刚1, 许职1, 李延晖1   

  1. 1. 华中师范大学信息管理学院, 湖北 武汉 430079;
    2. 华南农业大学数学与信息学院, 广东 广州 510642
  • 收稿日期:2013-11-15 修回日期:2014-02-02 出版日期:2015-04-20 发布日期:2015-04-24
  • 作者简介:张大斌(1969-), 男(汉族), 湖北潜江人, 华中师范大学信息管理学院, 教授, 博士, 研究方向:经济与社会预测预警、商务智能, 信息系统等.
  • 基金资助:

    湖北省自然科学基金创新群体项目(2011CDA116);国家自然科学基金资助项目(70971052);华中师范大学自主科研资助项目(CCNU14Z02016)

Study on Credit Risk Assessment Model Based on Automatic Clustering Using an Differential Evolution Algorithm

ZHANG Da-bin1,2, ZHOU Zhi-gang1, XU Zhi1, LI Yan-hui1   

  1. 1. School of Information Management, Central China Normal University, Wuhan 430079, China;
    2. College of Mathematics and Information, South China Agricultural University, Guangzhou 510642, China
  • Received:2013-11-15 Revised:2014-02-02 Online:2015-04-20 Published:2015-04-24

摘要: 随着风险评价的日益复杂化, 多维度、多时序等不规则的样本数据增加了评估的难度。本文建立信用风险评价的差分进化自动聚类模型, 并将其应用到我国上市公司信用风险评价中。该模型不要求事先知道分类的数据, 相反, 通过群体智能去寻找最优的分区。通过数据仿真, 并与遗传算法、决策树、BP神经网络模型进行信用风险评价的实证对比研究, 结果表明, 该模型能够非常准确的找到数据对应的分区, 大大提高了信用评估的准确性, 降低了风险成本, 对信用风险的管理和控制具有很高的利用价值。

关键词: 差分进化, 启发式搜索, 群体智能, 信用风险

Abstract: With the increasing complexity of risk assessment, multi-dimensional, multi-timing and other irregular sample data increases the difficulty of the assessment. In this paper, the establishment of credit risk evaluation of differential evolution automatic clustering model is applyed to our assessment of the credit risk of listed companies. The prior knowledge of classified data is not required in this model, on the contrary, swarm intelligence is used to find the optimal partition. By data simulation and empirical comparative study of credit risk assessment and genetic algorithms, decision tree, BP neural network model, the results show that the model can be very accurately to find the corresponding data partition, which greatly improving the accuracy of the credit assessment, reducing the cost of risk, making a high value of credit risk management and control.

Key words: differential evolution, heuristic search, swarm intelligence, credit risk

中图分类号: