DataScoring: Retail lending is one of the most popular and prioritized businesses in financial industry as well as demanding the most attention. Lending to potentially bad borrowers may substantially harm bank or credit union therefore this process must be addressed systematically by setting up automated and effective borrowers scoring process.
This problem is solved by our product:
1. We effectively score borrowers using big data.
2. We retrieve additional statistical data to conduct further communications with existing borrowers.
3. Optimize credit portfolio to minimize payment overdues and defaults.
We stack Microsoft technologies in production of the product - .Net, Azure Cloud, C# and CUDA.
Our algorithms and models are built upon (1) group of self-learning neuron networks, (2) system of input data normalization and semantic analyzer for text inputs; (3) customer psychological image design; (4) data clustering; (5) vanilla scoring systems.