Credit scoring model based on multilayer perceptron

The research establishes a neural network credit scoring model based on the principle of multilayer perceptron (MLP). It is used to evaluate the credit of 96 listed companies. The listed companies are divided into three groups: a "good" group, a "middle" group and a "bad" group according to their management situations. 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. All data come from the annals of listed company in 2000. The simulation results show that the classification correction rate of the neural network credit scoring model is 79.17%. In addition, we give the learning algorithm and step of the model in detail and analyse the experimental results.