Banks face rising delinquency risk as economic uncertainty puts pressure on household finances. Traditional collections processes are reactive, intervening only after missed payments. This engine aims to detect early signs of financial stress and trigger proactive outreach 2-4 weeks before delinquency occurs.
- Real-time Pattern Analysis: Detects signals like late utility payments, gambling spikes, or cash hoarding.
- Predictive Risk Scoring: Forecasts likelihood of default 2-4 weeks in advance using ML.
- Proactive Intervention: Triggers tailored outreach to preserve customer relationships.
- Frontend: Next.js, TailwindCSS, Shadcn UI
- Backend: Python, FastAPI
- ML Engine: XGBoost, Scikit-learn, Pandas
- Cloud: AWS Free Tier (DynamoDB, S3)
- Data: Synthetic Transaction Generator (Berka Dataset Base)
- Salary Shifts: Credited later than usual.
- Savings Drawdown: Balance declining week-over-week.
- High-Risk Spend: Increased gambling or lottery spend.
- Cash Hoarding: Increased ATM withdrawals.
- Utility Delays: Payments happening later in the billing cycle.