Research On Default Risk Identification Of Online Loan Based On Machine Learning Hybirdmodel

The Internet credit loan business is quite different from the traditional offline credit loan business carried out by banks and other financial institutions.The Internet credit loan business is generally applied based on the mobile terminal or Web terminal.The applicant only needs to provide simple personal data,and does not need to provide mortgage or collateral.Financial institutions can not conduct face-to-face examination on the applicant,but can only submit it through the applicant The information submitted,the credit report of the people's Bank of China and the data of the three parties are used to evaluate the default risk of the applicant.Due to the great differences in Internet technology and data application among Chinese banks and other financial institutions,some large commercial banks and licensed consumer finance companies can effectively identify the default risk of applicants through data mining and modeling technology.However,some urban commercial banks and rural commercial banks are still in the initial stage of exploration for the technology application of Internet credit loan business,and the ability of default risk management is weak.The larger the scale and the stronger the compliance of financial institutions,the more likely they are to adopt the logistic regression model to identify the default risk,because large financial institutions often need to face the inspection of the regulatory authorities,and need to explain the default risk identification model to the regulatory authorities.The logistic regression model is equivalent to a transparent box with strong explanatory power.While other machine learning models have better discrimination effect than logistic regression model,because the modeling process is similar to a black box,it is difficult to explain to regulatory authorities,so it is less used.Paipai loan is one of the earliest Internet credit institutions in China.It has complete Internet credit loan process management system,perfect risk management organization structure and rigorous risk control process,and uses big data and innovative technology to identify loan applicants with high credit default risk.This paper studies the risk management and control of paipaipai loan Internet credit loan,and provides a reference for the institutions to carry out Internet credit loan.At the same time,this paper also use the sample of the borrowers of the paipaipai loan,Python and SPSS as tools to build the logistic regression model,random forest model and the hybirdmodel based on these two models.The results show that the hybirdmodel has better discrimination effect than the logistic regression model;at the same time,the prediction ability of the hybirdmodel is similar to that of the random forest model,but the interpretation is better and more stable.At the same time,the calculation time of the model is shorter,which is more suitable for domestic banks,licensed consumer finance companies and Internet small loan companies.