Commercial Bank Performance Evaluation Based On Optimized BP Neural Network Model

Nowadays,under the development of economic globalization and financial liberalization,Chinese commercial banks are facing more changing and competitive situations and encountering more challenges.Therefore,it is of great significance to evaluate and analyze the performance of commercial banks.For commercial banks themselves,evaluation can help them find out their problems and improve their competitiveness.Meanwhile,this evaluation is also beneficial to all stakeholders of commercial banks.For example,reasonable performance evaluation results are more valuable for investors.Previous research on performance evaluation of commercial banks uses methods such as the balanced scorecard method which determines the weight of each index assignment mostly based on subjective past experience or expert advice.However,in this article,the evaluation method we use is the optimized BP neural network model.Because of the constant training process of this model,it can to some extent avoid the effect of subjective factor.Also,previous researches indicate that traditional BP neural network model can effectively evaluate the performance of enterprises.So,it is of both theoretical and practical significance to use this model to evaluate and analyze the performance of commercial banks.Based on previous studies on performance evaluation of commercial banks,this research focuses on 16 commercial banks in China.Considering the characteristics of different commercial banks,the research establishes the evaluation system based on 12 financial indicators which respectively showing the solvency,profitability,development financing ability,operating ability and security of enterprises.Firstly,we processed the original data with factor analysis method and then got the annual performance score of each commercial bank from 2013 to 2017.Secondly,we established an appropriate traditional BP neural network model,based on 12 indicators,and got the performance score as the target output values by factor analysis method.We also randomly selected part of the data to train the model and used this model to predict the rest of the sampledata.Finally,we got the mean square error(MSE)between output value and predictive value.Then on the basis of the traditional BP neural network model,we introduced the dynamic vector and the regularized function in L2 norm to optimize the model and used the optimized BP neural network model to study the sample data in order to avoid the situation that the convergence process of error rate in BP neural network model may fluctuate and excessive fitting may arise in the process of training.Finally,we compare the results of two models and find out that the optimized BP neural network model has a smoother convergence process and smaller mean square error rate than the traditional BP neural network model in the performance evaluation of commercial banks,which verifies the superiority and feasibility of the model.