Application Of XGBOOST Model In Credit Scoring Card

In the credit scenario,the user's credit rating has always been a key link.The traditional credit rating generally uses the logistic regression model to establish a credit scoring card to score the user's credit.Logical regression model has the advantages of simple model,easy to display and good classification effect.However,because logistic regression is a linear model,the non-linear factors contained in the information provided by users may not be fully extracted.That is to say,on the basis of logistic regression prediction,we still have some room for improvement.With the rise of the Internet,the competition in the credit industry has become particularly fierce.How to more accurately pass through good users is a concern of the credit industry.With the increasing heat of machine learning,using machine learning classification model to solve this problem has become a hot direction.In this paper,we use XGBOOST model,which is the most effective model to solve the classification problem.In Chapter 1,we will introduce the knowledge of credit scorecard.In Chapter 2,we will introduce the principle of traditional logistic regression model.In Chapter 3,we will introduce the principle of XGBOOST model.In Chapter 4,we will explain how to evaluate the model's superiority.Bad;in the last one,we will do the actual measurement of the data.Finally,we find that the prediction effect of XGBOOST model is better than that of logistic regression model.However,XGBOOST model can output a simple credit score card unlike logistic regression model.Therefore,in practical application,we first use the credit scoring card established by the traditional logistic regression model to score the users who need to be evaluated.For those who have lower scores and are refused loans,we select those who have higher scores and use the XGBOOST model to further determine whether there is default risk.