Big data risk control scheme?

It is generally divided into two parts: credit reporting, big data mining and risk control operation:

Credit big data mining:

Data related to risk control in massive Internet data.

Big data of e-commerce websites: Ali, JD.COM, Suning, etc. ;

Credit card website big data: I love cards, silver rate cards, etc. ;

Big data of social networking sites: Sina Weibo, Tencent WeChat, etc.

Small loan website big data: everyone's loan, credit information treasure, etc.

Payment website big data: Epro, Tenpay, etc.

Big data of life service websites: Ping An Zhang Yitong, etc. ...

Before data processing, it is very important to understand the business and data, which determines which data materials to choose for data mining. The workload before entering the "data factory" usually accounts for more than 60% of the whole process.

In terms of data raw materials, more and more Internet online dynamic big data are added. For example, by analyzing traces of network behavior, we can identify false information of loan applicants, while real Internet users always leave clues online. The timeliness of useful data for credit investigation is also critical. The effective dynamic data generally recognized by the credit investigation industry is usually data that goes back 24 months from now.

By obtaining multi-channel raw data of big data and analyzing it with mathematical operation and statistical model, the credit risk of borrowers can be evaluated. A typical domestic enterprise is the Shenzhou Rongda data risk control platform. Risk control with big data analysis is the core technology of Yiborui. Their raw data sources are very extensive.

The core technology and secret of their data factory is a variety of analysis models developed by them based on learning machines, which analyze the original information data of each credit applicant in more than 3000+ dimensions and get indicators that can measure their behavior, and this process can be completed within 5 seconds.

Risk control operation:

Pre-lending marketing: 1, existing customer development, new customer development; 2. Pre-approval and application score 3. Pre-approval, customer visit and pre-credit line evaluation.

Loan approval: 1, fraud screening, anti-fraud monitoring; 2. Apply for reclassification; 3. Credit approval. 4. Loan pricing.

Post-loan management: 1, behavior scoring model; 2. Quota management; 3. Risk early warning and advance collection; 4. Collection scores and collection strategies.

At present, the online speed of loan approval has achieved a breakthrough, and the loan approval rate has also been significantly improved. For the same type of users, the loan approval rate is about 15% by using extensive traditional risk control methods such as collateral and income proof, while the approval rate can reach more than 30% by using big data model combined with labor. As for the loans overdue rate, taking the default risk of 12 months as an example, the overdue rate of users screened by the China online loan approval model is half lower than that of users not screened.

Shenzhou Rong is the first Internet finance enterprise to make efforts in the big data risk control system. At the same time, some P2P online lending platforms under Sesame Credit and Ant Financial have begun to develop big data credit evaluation models.