data-storm-v1

This repo contains the code for Data Storm v1.0 - Team Extortion.

1. Use case definition - A default can occur when a borrower is unable to make timely payments, misses payments, or avoids/stops making payments.
2. Business problem - As a result of credit card default, bank accounts close, non-performing loans and bad debts get written-off. Therefore, the bank needs help in predicting and preventing credit card default to improve their Bottom line.
3. Key question(s) to answer - Which priority clients have the highest risk of credit card default?
4. How will analytics help?
   • Identify the lead indicators of credit card default. (eg: unknown missed payments)
   • Based on these attributes, generate a list of priority clients with high probability of credit card defaults.
   • For these likely to default clients, develop a list of marketing interventions to reduce credit card default.

Key Approaches

• Ensemble Model of XGBoost with CatBoost, which produced an F1-Score of 0.82466 on the Private Leaderboard.

Team Members

Shayen Yatagama Vinura Perera Pasindu Thisaranga