🔍 Preventing Credit Card Default Cases

We modelled ICP (Ideal Customer Profiling), NPL Prediction and Predicted very accurately which customers are most probable to default represents significant business opportunity.

We modelled both CWP, NPL Prediction to forecast Credit Card Defaulter using complex orchestration of our customized algorithms. These models learned Defaulters’ Patterns from a large amount of historical data including Card Holder’s Demography, Product details and Transaction history. This customized and configurable model started with a prediction accuracy of 83-87%, in 6 months with the continuous and adaptive learning methods this solution crossed the accuracy level of 90%.

With preventive actions executing on right time, it reduced defaulter by 7% in one year (to be more accurate in 6 months). With this specific success of AI (through an ad-hoc p) we stared our Discovery AI/CI.