Title: Forecasting global oil demand: applying machine learning methods
Authors: Ammar Neamah Awdah
Volume: 10
Issue: 3
Pages: 1-7
Publication Date: 2026/03/28
Abstract:
The dynamic instability of world energy markets and the growing role of the technological transformation process require the search for true and effective tools for forecasting demand for raw materials. This paper proposes and justifies a hybrid neural network architecture that combines the strengths of convolutional layers for spatial feature extraction, recurrent layers for long-term dependency calculation, and a temporal attention mechanism for adaptive weighting of market shocks, which together provide a reliable forecast of global oil demand. The scientific significance of the research lies in the development of a comprehensive forecasting method that can effectively analyze nonlinear relationships between macroeconomic indicators and structural changes in energy consumption.The results of the testing confirm the high robustness of the proposed approach, which provides a significant increase in the accuracy of approximation and the reliability of forecast values compared to traditional econometric methods and basic machine learning algorithms.