Credit risk management systems have transformed from the focusing on the total loan process life cycle in the first generation, to the application of structure data such as linear scoring card in the second generation. The new generation of credit risk management systems in the big data era are of exemplary reliability with:
1.The ability to process both structured and unstructured data simultaneously;
2. The ability to model risk and behaviour using non-linear models.The following figure illustrates the main component of the third-generation big data credit risk management system: Big Data =AI + BI + Modelling
There are 3 conditions to meet for a robust and reliable AI platform:
1. Legitimate & reliable data sources - Relying solely on bank’s internal data and credit data from the central bank is no longer sufficient as related external unstructured data that does not
involve privacy issues will further enhance accuracy.
2. Experienced data scientists – Build a strong AI platform to deal with massive unstructured external data with deep network training techniques.
3. Subject matter experts – This is often an overlooked point. For platforms to function reliably, subject matter experts must be able to transfer their field knowledge correctly to the data scientists to write out correct AI algorithms. However, it is even harder to find such experts compared to experienced AI scientists.
With a powerful artificial intelligence platform to unify internal and external heterogeneous data, the new generation of risk management system that covers the complete loan life cycle is developed.