The Practice Study of Consumer Credit Risk Based on Random Forest

Abstract

How to evaluate and identify the potential default risk of the borrower before issuing the loan is the basis and important link of the credit risk management of modern financial institutions. Based on the data provided by an auto finance institution, This paper mainly studies how to analyze the historical loan data of auto financial institutions with the help of the idea of unbalanced data classification, and predicts the possibility of loan default based on Random forest classification model, which provides a reference for the risk control of this institution. Introduction According to the data of China auto industry association, the sales volume of China's auto market in 2015 was 24.597.76 million units, an increase of 4.7% year on year, is the lowest growth rate since 2012. On the contrary, the growth rate of auto finance business has maintained a high level. Relevant data show that in 2014, the size of auto financial market exceeded 700 billion, and the penetration rate of auto finance exceeded 20%. In 2015, the overall size of China's auto financial market was about 800 to 900 billion, and the overall penetration rate was about 35%. Figure 1. Demostic auto market trend Introduction to Credit Risk in Auto Finance At present, in the credit risk management of auto finance companies, subjective judgment is the main way to identify and evaluate the risk, which means based on experience and full of randomness. The basic data used in the model mostly come from the qualitative judgment of credit personnel, which cannot achieve the ideal effect of risk management. In future business operation, in order to improve the technical level of credit risk management, most auto financial institutions willing focus on quantitative indicators, establish a risk control mechanism using loan risk degree model and behavioral scoring model as tools, and use mathematical statistics model to measure and analyze risks, so as to achieve a reasonable offset of risks. Under this context, this paper provides a reference for credit granting of auto financial institutions by data modeling of an auto financial company.