摘要
在确立了商业银行信用风险评价指标体系的基础上,建立了基于模糊神经网络的商业银行信用风险评估模型.该网络具有四个因子输入,一个衡量商业银行信用风险的输出,总共六层的结构,且模糊规则层具有根据具体问题情况进行调节的能力,优于神经网络完全黑箱操作的特点.利用Matlab6.1对167组样本数据进行实证分析,训练结果表明网络预测误差小.
A commercial bank credit risk assessment model based on fuzzy neural network is established using the credit assessment index system established for commercial banks. This network is a 6-storey structure with 4 factor inputs and one output measuring the credit risk of commercial banks. The fuzzy rule layer has the capability of making necessary adjustments in accordance with specific conditions of problems. The operation of this model is much better than the totally black-box operation of a neural system. A substantiation analysis has been made with 167 groups of sampled data using Matlab 6.1, and training results indicate that the network prediction has less error.
出处
《系统工程理论与实践》
EI
CSCD
北大核心
2004年第11期1-8,共8页
Systems Engineering-Theory & Practice
基金
国家社会科学基金(02BJY126)
黑龙江省青年科学基金(QC04C25)
关键词
商业银行
信用风险评估
模糊神经网络
实证研究
commercial bank
credit risk assessment
fuzzy neural network
substantiation research
