Research On Credit Evaluation Based On Device Fingerprint Behavior

With the development of economic and the increasing demand for credit,credit card has become an important tool of microfinance.The demand for credit cards has been increasing year by year,and the number of cards issued in 2018 has reached 0.97 billion.At the same time,the outstanding balance of credit cards reached 6.85 trillion Yuan at the end of the same year.Illegal cash out and malicious overdraft cases show the characteristics of sudden,diverse and cross regional.Some lawbreakers use the Internet to speed up the difficulty and scope of fraud,and use high-tech means to explore audit and examine loopholes to get credit cards.For this reason,the banks have to bear huge losses every year.As mentioned in the Basel Accord,credit risk is the most important risk faced by banks.As the first line of defense for credit risk management,credit evaluation has a significant impact on its subsequent risk control.With the development of science and technology,the fraud methods are constantly updated and iterated,and various new fraud methods make it difficult for banks to parry,such as agency,group case,false name,etc.Traditional expert rule-based risk control requires a lot of labor and time,the credit card center can only be passive anti-blocking.If the fraud characteristics are not obvious or the large-scale scattered parts,it is difficult to find the risk.The traditional credit evaluation method has been difficult to meet the needs of the current business.The new business type needs not only the combination of internal and external data,but also the credit evaluation tool that can actively find fraud and prevent similar risks.As a new technology,device fingerprint has been used in the field of risk control and credit evaluation.Device fingerprint refers to the unique and non-repetitive device identification based on multiple characteristics of devices and media generation.As an identification technology,it is a very mature technology in the network world.By collecting the non-sensitive device feature information of the end-user environment,a unique user identity mark is formed in the background of the Credit Card Center.Through this mark,the owner's behavior track can be seen and marked.If there is risk,when he re-enters,the audit end can timely find the risk,locate the risk subject,actively remind the audit personnel,and put the stock customer Associate with non-cardholder,integrate relationship data and build customer relationship network,and associate all suspicious users.As a new credit evaluation tool,equipment fingerprint provides auditors with a way to describe the potential credit risk fraudsters' portraits and show which behavior patterns may exist.This is difficult to achieve when information technology is underdeveloped in the past.With the development of science and technology,commercial banks gradually attach importance to big data.Equipment fingerprint is the development trend of risk control in the future.It is at the forefront in both academic field and practical application,and plays an auxiliary role in the field of credit evaluation and anti-fraud.With the development of science and technology in the future,the application scope of equipment fingerprint will be more extensive and the accuracy will be higher.It will provide the idea of counterfeiting audit for the future anti-fraud and credit evaluation work,and provide solutions for risk prevention and control.Due to the lag in the issuance of credit card,the user application feedback data of a commercial bank's Credit Card Center from the third quarter of 2018 to the fourth quarter of 2018 is retrieved,and the original audit system without using the device fingerprint and the new audit system using the device fingerprint logic are compared to see whether the two systems have an impact on the subsequent counterfeit rate and the reverse rate of credit card.Through the new system,cases that can directly confirm the risk will directly enter the return process,and those that cannot directly confirm the risk or are in the gray area will be transferred to manual processing;the auditor will judge whether the applicant has credit risk according to the risk information associated with the equipment fingerprint.This paper studies the background of credit card credit evaluation,analyzes and summarizes the credit card pre loan credit evaluation methods and the impact of pre loan fraud,analyzes the development process of each stage of credit card pre loan audit,combs the internal and external data sources of credit card and summarizes and analyzes its shortcomings,and puts forward the improvement scheme of credit card credit evaluation,that is,based on the original audit system,add "design" to the credit card pre loan audit system.The "fingerprint backup" logic,through the "equipment fingerprint" left by the applicant in the network application,compares with the known risk cases to determine whether the customer has credit risk.According to the feedback data collected by the backstage in 2019,the acceleration rate of M1 + in all channels decreased in 2018,down nearly 50% month on month.The overall FA incidence rate in the current quarter is 3.1bp,down nearly 46% compared with the previous quarter,which has reached the lowest level since the third quarter of the previous year,and the risk level of the new system channel has declined significantly.Through the correlation of equipment fingerprints,the case digging rate and reporting rate increased significantly,while reducing the FA incidence rate did not affect the case rate.It is found that the "device fingerprint" logic can play a very good role in group case discovery,fraud early warning and other aspects.For organizations such as credit card centers with higher security requirements,although they currently read fingerprint information of non-sensitive equipment,have limited awareness of abnormal behaviors,and still need to judge risks manually twice,and also have certain requirements for the number of outlets and the configuration of examine equipment,they can still use this logic to find changes in the user environment and equipment of applicants,so as to prevent mass agency and group case happen.