This ArXiv paper proposes an AI/ML framework to predict invoice dilution in supply chain finance, mitigating non-credit risks and margin losses. It employs leakage-free two-stage XGBoost, Kolmogorov-Arnold Networks (KAN), and ensemble models trained on production data across nine transaction fields. The method supports real-time dynamic credit limits, reducing reliance on buyer's irrevocable payment undertakings (IPU).
Key Points
- 1.Introduces leakage-free two-stage XGBoost for accurate dilution forecasting
- 2.Integrates KAN for advanced non-linear modeling in finance
- 3.Uses ensemble models to boost prediction robustness
- 4.Evaluated on real production data from nine key transaction fields
- 5.Supplements deterministic algorithms for dynamic credit limits
Impact Analysis
Enhances supply chain finance adoption for sub-investment grade buyers by replacing IPUs with data-driven predictions. Reduces margin erosion from dilutions, benefiting fintech AI applications in risk management.
Technical Details
Framework prevents target leakage via two-stage training on historical data. Combines XGBoost's gradient boosting with KAN's interpretable splines and ensembles for superior performance over baselines. Processes buyer-supplier pairs across nine fields like invoice amount and payment history.