| In 2008 after the financial crisis,China's rapid development of shadow banks,shadow banking business risk also showed a rising trend.Therefore,the forecasting model of the risk of shadow banking is not only beneficial to the commercial banks to carry out risk management,but also to maintain the stability of the whole financial system.Based on the research of domestic and foreign scholars,this paper chooses 11 indicators of 114 commercial banks to predict the risk of China's commercial banks engaged in shadow banking business.The sequence of the data is tested for the smoothness of the data.Because the regression analysis has a very low degree of goodness of fit,we use the stepwise regression model to eliminate five non-significant independent variables,leaving five independent variables for stepwise regression,but the prediction error of stepwise regression is 20.58% and the error is large.So it is predicted by BP neural network.The degree of fit is 76.97% and the prediction error is 10.86%.The prediction effect is general.This may be due to the influence of weight and threshold randomness,so the genetic algorithm is used to optimize the BP neural network.When the genetic algorithm is used to optimize the parameters,the maximum genetic algebra size and the number of individuals have a significant effect on the optimization performance of the algorithm.By trying to obtain the genetic algebra selection 62,the selection of the number of individuals is the best,and the training network The test set is 96.07%,the prediction error is 0.84%,much better than the optimization.When the predicted risk value is greater than the average 0.39 of the risk value,the bank should be vigilant and strengthen risk prevention.At the end of the article,the conclusion of this paper is consistent with the current part of the law and make some suggestions.In general,the BP neural network optimized by genetic algorithm can be used to predict the risk value. |
PDF Full Text Request