How to use big data to conduct financial risk control for JD, Zhangzhong and Bairong
Said that Jingdong's financial resources are mainly obtained from e-commerce data systems, a large number of investment data obtained through out of system cooperation, and some from a large number of model variables and multidimensional data.
At present, under the guidance of this concept, financial services for 100 million users, Hong Kong vps More than ten thousand enterprise customers, in the past, there were nearly one trillion transactions, and the capital scale has increased in the past three years.
Zhang Jinghua: Data stream forms value stream
Zhang Jinghua, CEO of Palm Public Services, believes that the core competence of the big data control team is to have a large amount of data, real-time decision-making, and data before, after and after the loan. Strong data collection and processing capabilities not only include maintaining data stability, but also identifying third-party fraud data, etc. On how to improve the technology of data processing, Zhang Jinghua said that a lot of technical work is needed in how to wash the clothes, share the processing and distributed computing, and make risk control decisions.
Zhang Jinghua, taking "user sensitive information" as an example, pointed out how this technology can protect user privacy, enable risk control personnel and employees of different customers to use and support the system, and make a cycle. Intelligent decision engine practice The whole business is to use public payment services to make decision engines, including credit risk, which allows users to obtain loans in seconds; Smart payment, WeChat, through the collection of experience and other users in different situations; This is precision marketing, which aims at different users to achieve loan needs. This decision model can realize automatic learning and feeding system, which is aimed at the public and is realized through artificial intelligence machine learning.
Zhang Shaofeng: It is an auxiliary variable to make credit more accurate
Zhang Shaofeng takes Home Credit, a commercial customer with big challenges, as an example. This article describes the importance of big data and deep learning technology. Zhang Shaofeng, the largest consumer finance company of Home Credit in China, has a loan of billion yuan this year. To identify fraud risks and identify credit risks, Home Credit is an urgent problem to solve. Zhang Shaofeng said that these two aspects are involved in the comparison of loan application equipment fraud prevention, blacklist filtering, identity authentication, network abnormal behavior and application information.
Home Credit's customer base has sunk, and its customers are mainly blue collar workers and migrant workers, Hong Kong vps But these people have almost no regular credit data, and can't find the bank's credit report in 80 or 90 points. Zhang Shaofeng said that through the algorithm and machine learning of big data technology, after six months of exploration, the effectiveness of the financial services model from the original to. Taking the calculation of repayment capacity as an example, we first need to understand the income of users, the second is liabilities, and the third is liabilities
