Machine learning-based prediction of global solid waste generation and composition.

The rapid growth of municipal solid waste (MSW) driven by urbanization, population expansion, and economic development has emerged as a critical global environmental challenge, thereby requiring effective planning and resource management. However, accurate prediction of MSW generation and composition-which is essential for effective governance-remains hindered by inconsistent and highly heterogeneous global data. To address this gap, this study proposes a data-driven framework using multi-linear regression (MLR) and artificial neural networks (ANN) to forecast MSW across 217 countries. These models incorporate socioeconomic and demographic parameters, including GDP, population, literacy rate, urbanization, and household size. The results indicated that ANN outperformed MLR in terms of predictive accuracy, achieving R² of 0.94 for total MSW generation, compared to ~ 0.57 for MLR and ~ 0.68 reported in the existing global model. While prediction of waste generation exhibited strong accuracy, composition prediction remained challenging (R² up to 0.15), indicating the influence of unaccounted behavioral and regional factors. Despite limitations in accuracy due to data heterogeneity and compositional complexity, the findings of this study can support policymakers and planners in enhancing waste management strategies, optimizing resource recovery, and enabling data-driven decision-making toward sustainable and circular waste management systems aligned with the Sustainable Development Goals.