Commercial Credit Difference Evaluation and Prediction Model: Based on Neural Network

Abstract

The different capital configurations of bank accounts for different credit tendency which will affect the currency condition as well as whole finance system. The mercurial and nonlinear factors of economy usually brought the difficulty when predicts. This study adopts the feed-forward backprop network (BP), conceiving predicting modeling with the sample of various banks’. The simulation results indicated that the model performs well in anti-interference and accurate in prediction error (less than 2%). Moreover, we got the result that non state-own banks tend to be more cautious than state-own by 10% on average.