Дослідження методів інтеграції моделей машинного навчання у розподілені вебсистеми електронної комерції для прогнозування попиту на товари

Abstract

The work is focused on researching methods for integrating machine learning models into distributed e-commerce web systems to forecast product demand. It examines the architectural approaches for deploying ML models within a microservices infrastructure, ensuring high scalability and low latency. The research highlights the importance of data preprocessing and real-time synchronization between distributed components. The integration of forecasting models allows e-commerce platforms to optimize inventory management, reduce operational costs, and improve customer satisfaction. The study also discusses the selection of appropriate frameworks and communication protocols between the web system and the ML inference engine. Special attention is paid to Clean Architecture and design patterns like Repository and Dependency Injection to ensure system flexibility and testability.