V-fold交叉验证和BP神经网络在信用评价中的应用

An Application of V-fold Cross-validation Technique and BP Neural Networks to Credit Analysis

研究关于公司神经网络信用评估问题的现状,提出一套甄选方法准则,用于建立适合于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BP回归神经网络的信用评估模型;采用V-fold交叉验证技术,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。

We study the present situation about credit evaluation of corporation neural network. Some methods and guide lines are put forward, so that some distinguishing indexes can be chosen. Thus credit-score models applied to Chinese corporations are set up by BP neural networks. Finally, an experiential research of the credit-score models＇ scoring powers is carried out using actual index data by V-fold Cross-validation technique.