International Journal of Management

The complexity of forecasting global demand has increased due to volatile geopolitical conditions, pandemic disruptions, climate shocks, and changing consumer patterns. Traditional econometric models often fail to capture nonlinear relationships, multi-source data streams, and dynamic causal interactions. Machine learning (ML) has emerged as a powerful alternative for predictive analytics, enabling more accurate scenario modeling using high-dimensional data, real-time updates, and cross-market integration. This paper evaluates ML-based forecasting approaches for global demand, comparing statistical, hybrid, and deep-learning models. The study proposes a new Global Demand Forecasting ML Framework (G-DFMLF), integrating feature engineering, multi-modal data fusion, cloud-based pipelines, and explainability protocols. Applications span energy markets, consumer goods, agriculture, pharmaceuticals, and automotive industries.