基于神经网络的企业信用评估模型

Enterprise Credit Rating Model Based on Neural Networks

在分析信用评估重要性和信用评估国内外现状的基础上,提出了用于企业信用评估的指标体系,给出了指标体系中离散数据的量化方法和指标数据的归一处理方法。根据这一指标体系建立了基于BP神经网络的企业信用3层神经网络评估模型。该模型通过给定的教师数据训练后,具有了对企业信用的预测功能。实验结果证明,该模型用于企业信用评估,减少了企业信用评估传统的定性方法中权重确定的人为因素,评估正确率达到了92.2%。

On the basis of analyzing the importance and actuality of credit rating in the world, the index system used for enterprise credit rating is proposed and the method of quantifying the scatter data and the method of normalizing all the data in the index system are given. Subsequently, a 3-layer neural network enterprise credit rating model based on the index system and neural networks is built. After being trained by the given credit data of the enterprise, the model can be used for rating the enterprise credit. Empirical results show that this model can decrease the personal factors and increase the veracity and authority and the correct rate is 92.2%.