Research On Personal Score Model In Consumer Finance Industry

China carried out consumer finance company pilot since 2009.With the increase in spending power,consumer attitudes change,credit consumption,ahead of consumption and other models gradually accepted by the public,Consumer financial market has grown China’s credit business has been high-speed development.However,with the development of credit business,the traditional line of manual approval of customer needs has been unable to meet the growing customer base,There is a big gap between the booming consumer finance business and the domestic backward risk management system.In the face of the foreseeable huge development opportunities,we can also see the problems we encounter and bottlenecks.At present in the country,whether it is the choice of personal credit risk index system or the establishment of the model did not form a unified standard,Each financial institution has its own credit evaluation system.At present,the method of personal credit scoring model is divided into statistical method and non statistical method,the research focus lies in the accuracy and stability of the model.This paper analyzes the research results of personal credit risk assessment at home and abroad.Combine the sample data,Based on the AdaBoost algorithm,three single personal credit risk models are established and compare the predictive effects of each model.the forecast of decision tree model is very good,and the accuracy is as high as 90%.The prediction accuracy of the Na?ve Bayes model and the Support Vector Machine is equivalent,But the stability of the Naive Bayesian model is better than that of the support vector machine,Support Vector Machine in the second category of false rate is higher.Therefore,this paper argues that it is a better method to establish credit rating model based on decision tree algorithm under AdaBoost framework.