摘要
BP神经网络用作信用等级分类可取得较好的效果,但在过分要求输出信用分值时效果不佳。针对该缺陷,本文采用自适应神经网络(ANFIS)和Elman网络研究公司信用评分。文中提出了一套甄选方法准则,用于建立适合我国企业的信用评分指标体系;然后依据该指标体系建立了基于Elman网络和ANFIS的信用评估模型;采用V foldCross validation技巧,利用样本公司实际指标数据对该模型的评分效果进行了实证研究。
Although strong power is given by BP neural networks to discriminate credit grades,it behaves bad when scoring credit conditions.Then the Adaptive Neural Fuzzy Inference System and Elman neural networks are proposed to score corporate credit conditions in this study.Some methods and guide lines are put forward,so that some distinguished indexes can be chosen.Thus credit scoring models applied to Chinese corporations are set up by Adaptive Neural Fuzzy Inference System(ANFIS) and BP neural networks.Finally,experiential research of the models' scoring power is carried out by use of actual data by V-fold Cross-validation technique.
出处
《管理工程学报》
CSSCI
2005年第1期69-73,共5页
Journal of Industrial Engineering and Engineering Management
基金
国家863基金项目(2002AA41361)