The Research Of Personal Credit Risk Assessment Model Based On BP Neural Network

With the development of social economic,our country gradually tends to the "credit economy".Combined with the optimization of our country's economic policy adjustment,credit business has gradually sprung up in the financial market in China.In fact,with the rapid develop of urbanization,more people apply the "hose loan","car loan",or another personal credit business.However,the time of our country's credit system construction is late.We don't have a solid foundation and enough experience.There are many shortcomings in our credit system,especially in the aspect of credit risk controlling.An in-depth study shows,at present,most banks in China widely use marking system to evaluate individual credit risk for quite a long time.Moreover,the evaluation indicators don't keep pace with the times,because they don't have any change.Obviously,this assessment way is with a very strong subjectivity,and the evaluation indicators are relatively outdated,slightly rigid and simplex.Therefore,this paper puts forward a new personal credit risk assessment model based on BP neural network.It hopes to provide a positive reference significance for personal credit business to banks.This paper essentially needs to solve three problems: optimization of BP neural network algorithm,building a new evaluation indicators system and building an evaluation model:(1)Optimization of BP neural network algorithm: BP neural network is limited by the initial weights and thresholds,and easy to fall into the local minimum point.The hybrid frog leaping algorithm is a bionic intelligent optimization algorithm,which has good global search ability.Therefore,this paper proposes a solution to optimize the BP neural network with the improved hybrid leapfrog algorithm.(2)The evaluation indicators system: in according with the actual development of our country's credit system,this paper references the evaluation indicators of four mainstream personal credit evaluation methods both at home and abroad(include: China construction bank personal credit evaluation,credit sesame of ants financial,RenRenDai and American FICO).On the basis of traditional evaluation indicators,it eliminates some poor recognition personal credit evaluation indicators,and adds 2 another indicators about internet,just for they have a strong ability to predict the personal credit.The new evaluation indicators system completely broke the traditional which only consider the "offline indicators ",and realizes a comprehensive evaluation by using " online" plus "offline".(3)Evaluation model: Using the improved hybrid leapfrog algorithm to optimize the BP neural network model to establish the assessment model.Then it will do a deep forecast analysis with 18 indexes of the online and offline data to implement predict personal credit risk rating.In this way,it can avoid the subjectivity of traditional evaluation methods,and shorten the time of assessment flow,thus improve the efficiency of evaluation.Finally,the accuracy of the prediction model is proved by experiments.