多层感知器信用评模型及预警研究

本文利用多层感知器 ( MLP)原理建立神经网络信用评价模型 ,用来对我国 2 0 0 0年 1 0 6家上市公司进行信用评级 ,并进一步对我国 2 0 0 1年公布的 1 3家预亏公司进行预警研究 .按照各上市公司的经营状况分为“好”、“差”两类 ,每一类由 5 3家上市公司构成数据样本 .对于每一家上市公司 ,主要考虑其经营状况的四个财务指标 :每股收益 ,每股净资产 ,净资产收益率和每股现金流量 .仿真结果表明 ,本文所建立的神经网络信用评价模型有很高的分类准确率 ,达到 98.1 1 % .又由于该信用评价模型有很强的适应能力 ,故可以进一步用来对企业的财务危机进行预警研究 .预警实证分析表明 ,该信用评价模型对我国 2 0 0 1年公布的 1 3家预亏公司进行预警分析 ,预警准确率达到 1 0 0 % .此外 ,文中还给出 MLP网络模型的学习算法和步骤

Study on Miltilayer Perceptron Credit Scoring Model

The research establishes neural network credit scoring model by using miltilayer perceptron (MLP) principle. It is used to evaluate the credit of 106 listed companies of our country in 2000 and futher forecast 13 pre\|losted companies in 2001. The 106 listed companies are divided into two groups: a ″good″ group and a ″bad″ group according to their management situation. To each listed company, we consider four primary financial indexes: income per share, net asset per share, return rate of net asset, and cash flow per share. The simulation results show that the MLP credit scoring model has very high classification correct rate (98 11%) . And it has very strong adaptive ability, so it can be further used to predict financial distress of the companies. The forecasting research shows that the forecasting accuracy rate of the model is 100%. In addition, the paper is given the learning algorithm and step of the model in detail.