Brain will extend its existing proprietary Machine Learning and Natural Language Processing platform to create a product aimed at the early detection of Non Performing Loans for small and medium enterprises. The system will combine structured and unstructured financial data for for small and medium enterprises also including external data, for example economic and market conditions indicators.
The engine will be dynamic in the inclusion of features and will keep track of the most relevant ones, therefore creating an intelligible model as opposed to a ‘black-box’ system. The individuation of the most relevant factors contributing to the default probability and their dependence from external conditions will allow to create a dynamic scorecard for the evaluation of new applications for credit.
- Use of Model–view–controller (MVC) as a software design pattern to divide the software into three separate but interconnected elements (data model, user interface and computational model) to facilitate the interoperability of the appliction with external systems.
- Use of RestAPI protocol to standardize the backend communication with user interface and external systems