The Applications Of Reject Inference In Credit Scorecard Models

With the rapid development of China's economy,credit consumption has increasingly dominated the way of individual consumption.Small loan companies have mushroomed and developed rapidly.As of January 2018,there are currently 3,000 domestic companies.Small loan companies.Compared with the mature personal loan industry in Western countries,the disadvantage of China's industry is that China's credit system is still in its development stage.Each company has its own complete set of credit model.In the process of personal consumer credit development,the main problem faced by microfinance companies is to assess the individual's credit risk to determine whether to approve applicants' loan needs.Therefore,the study of personal credit scoring models has very important practical value.In the past,many scholars focused their research on the personal credit scores of bank data.With the increase in the business volume of small loan companies,personal credit loans are also accumulating.The research focus of this article is to use personal loan data to establish a personal credit scoring model,which is the risk control model.Using the knowledge of applied statistics and machine learning,the system restores the entire steps of the actual credit score card model.The statistical methods used to make credit scorecard models include logistic regression models,KNN,neural networks,decision trees,and random forest methods.In addition,non-statistical methods or rules can also be used to judge.This article mainly uses Logistic regression model to solve the problem of approval of personal credit requirements.There is a sample bias between the accepted sample of the applicant and the total number of applicants.The model used is a sample of accepted applicants.The model of the credit score created will have the problem of sample bias.In order to obtain more reasonable and accurate prediction results,it is possible to reduce this deviation as much as possible by refusing to infer the rejection of the sample.This paper gives a detailed introduction to the method of refusing inference,and makes an empirical analysis of the hard cut method,and compares it with a model that does not reject rejection,and draws a conclusion.