基于SOC网络的商业银行信用风险识别模型

针对自组织竞争(SOC)神经网络在解决模式分类问题上的优势,结合主成分分析法来构建商业银行信用风险识别模型.首先构造一套用于描述贷款企业信用状况的指标体系,然后使用主成分分析法提取特征指标,再采用SOC神经网络进行非监督分类.通过选取陕西省2007年度在沪、深两市交易的26家上市公司作为样本进行实证分析,实证结果表明:模型对信用风险具有较强的识别能力,同时对商业银行还有较好的预测功能.

A Credit Risk Recognition Model for Commercial Bank Based on SOC Neural Network

This paper constructs a credit risk recognition model for commercial banks by combining Self- Organ-ized Competition (SOC) neural network's superiority in pattern classification with principal component analysis. First, a set of index system is established for credit risk assosment,then extracting character index from principal componet analysis,finally finishing non-supervisory pauem classification by SOC neural network . At the same time, 26 listed companies of Shanxi are selected as samples in 2007. The results show that the model has strong risk recognition ca-pability and preferable risk predicting fanction for commercial banks.