Use case
The client wanted to build a solid credit risk methodology as the foundations of the business where micro-credits are given to suppliers that are considered reliable according to past transactional data of purchase orders and invoices.
Solution
A risk model has been developed considering different competing time series forecasting methods in backtesting with historical data. The model has been deployed in production by Twinco.
Besides that, a default tracker dashboard has been generated as a ShinyApp, so that history of suppliers can be explored and variations in supplier patterns can be detected as early warnings for near future defaults. The dashboard also contains useful aggregated KPIs at portfolio and market level.
