Research on the Construction of Personal Credit Score Model Based on WOE Analysis and Logistics Model

Abstract

As the concept of credit consumption enters people's lives, personal credit loans are gradually becoming a consumer demand, which makes the problem of credit risk in the bank lending business even more serious. To better protect the interests of financial institutions, investors and consumers, and to make the financial world more balanced and secure, scoring and modelling personal credit can reduce the likelihood of credit risk by identifying and quantifying risks in advance, reducing losses and making reasonable and effective loan plans. This has important implications for financial institutions, investors and consumers alike, and plays a very important role in economic development. In this paper, we will study the model construction of individual credit scores by obtaining data from customers' basic attributes, repayment ability, credit transactions, property status, loan attributes, other factors and time windows, processing and analysing them, and using WOE analysis to determine whether the indicators are economically meaningful, and correlation analysis to check the relevance of variables and IV screening variables. The logistic regression model was then converted to a standard scorecard format through WOE transformation and the model was then tested to obtain the scoring criteria.