Системний аналіз та прогнозування фінансових показників партнерських платформ електронної комерції методами машинного навчання

Abstract

The paper is devoted to the systems analysis and predictive modeling of financial metrics for e-commerce affiliate platform partners. The architecture of an intelligent information system is described, including a centralized database, an asynchronous backend framework, an isolated machine learning module, and a client web interface. Functional analytical modules for both platform users and administrators have been implemented. For time-series forecasting of financial inflows, a multiple linear regression method with additive decomposition into a linear trend and a weekly seasonal component using Fourier harmonics was applied. Experimental studies on real production data confirmed the high accuracy and computational stability of the proposed systems solution.