Empirical Analysis Of Personal Credit Evaluation Model

In recent years,with the gradual development and popularization of Internet communication and Internet,Internet finance industry is booming in China,and Internet finance loans are becoming a trend.Since 2007,when online loans entered China,this kind of credit trading model has gradually developed and gained popularity,and has been playing an important role in the Internet finance industry.But at the same time,the popularity of this by credit unsecured loans,credit risk has it,so it causes the loan platform need to be more strict audit application user credit information,however,due to domestic credit system is not perfect enough,and the risk of credit loans unsecured mode control research is not enough,lead to many loans overdue customers and bad debts rate increase platform,so that eventually face the risk of failure.With the mature of big data technology,net lending industry also gradually choice by analyzing the big data mining technology to the user data,find can help screening high quality customer data,in addition to able to take advantage of business experience,also can use the data model for scientific judgment customer qualification,thereby increasing profits and reducing the loss of industry.This article mainly through Lending Club P2 P Lending platform for the construction of the original data for subsequent model,due to poor data quality,a format chaos or missing values and outliers,and non equilibrium problems existing in the original data,so before the establishment of data model,need for preprocessing of the raw data in this article,through SMOTE algorithm is one of the few samples increased,and more of the value of continuous attributes and the recoding of the feature points and discrete variables,in order to improve the data quality,eventually improve the modeling efficiency.In addition,this paper,by using logistic regression model and random forest model to forecast evaluation of sample set,and by using ROC curve and confusion matrix to evaluate model,to get more effective data model,which provide the scientific reference for loan platform and reference.