Research On Credit Risk Of BP Neural Network Based On Improved Genetic Algorithm

We first derive the forward training of BP neural network input signal and the reverse propagation process of error signal,extend the weight and bias update of the typical three-layer network structure to the parameter update of any layer number,use genetic algorithm to optimize the initial weight and threshold of neural network,so that the network can learn the mapping relationship between input and output more quickly.Then through the principal component analysis of the selected evaluation index,the basic characteristics of the borrower,the borrower's willingness to repay,the borrower's repayment ability,the borrower's basic quality four main components as the input of the neural network,and the borrower's credit risk level as the output to establish the model,the prediction accuracy of the model is measured by the percentage of error between the real value and the predicted value.The experimental results show that BP neural network optimized by genetic algorithm outperforms standard BP neural network in terms of iteration number and training error,the accuracy of the standard BP neural network on the verification set is 67% and the accuracy of the optimized neural network is 83% if the Absolute error is 5%.Through the principle analysis and empirical research on neural network,it is shown that the neural network with three-layer structure has good predictive ability in the evaluation of credit risk level of borrowers in P2 P online loan industry,and can establish an efficient and practical credit risk rating model.