Совершенствование системы кредитного скоринга в банках с использованием технологий AI

This paper examines approaches to improving bank credit scoring systems through the application of artificial intelligence and machine learning technologies. It discusses changes in data sources, feature selection, model architecture and performance metrics, as well as requirements related to explainability, fairness and model risk management. Particular attention is given to the use of alternative and transactional data, ensemble tree models, neural networks and explainable AI techniques in modern credit risk assessment. The study also addresses regulatory expectations, model governance, transparency, stability over time and the need for continuous monitoring of AI-based scoring systems. The paper concludes that effective implementation of AI in credit scoring requires a balance between predictive accuracy, interpretability, fairness, operational controllability and regulatory compliance.

Hali tarjima qilinmagan