ARTIFICIAL INTELLIGENCE IN A FORECASTING AND ANALYTICAL ENTERPRISE MANAGEMENT SYSTEM

Abstract

The aim of this study is to develop a methodological framework for integrating artificial intelligence into a predictive analytics system for enterprise management. A closed-loop architecture is proposed that combines data preparation, forecasting, scenario analysis, decision optimization, and model drift control. Based on 141 months of observations, a seasonal model, exponential smoothing, and gradient boosting of decision trees are compared. It is established that increasing the complexity of the algorithm does not guarantee improved accuracy when the time series regime changes. The need for a competitive selection of models, taking into account the management loss function, is substantiated.