Hybrid Statistical-Machine Learning Framework for Forecasting Municipal Solid Waste Generation in Indian Cities: Seasonal, Climatic, and Event-Driven Dynamics

Rapid-urbanization, population-growth, and changing consumption-patterns have led to a continuous rise in municipal-solid-waste (MSW) generation across Indian cities, exerting substantial pressure on collection, treatment, and disposal infrastructure. Reliable forecasting of MSW is therefore essential for sustainable urban-planning, landfill-diversion, recycling-optimization, and WtE-deployment. However, forecasting-practices used by many urban-local-bodies continue to rely on linear extrapolation or single-model approaches that inadequately capture seasonal-variability, climatic-influences, and event-driven waste surges characteristic of Indian urban systems. To address these limitations, the present study proposes a hybrid statistical-machine learning framework exclusively focused on MSW-generation forecasting. The framework integrates interpretable statistical time-series models, namely ARIMA and Holt-Winters exponential-smoothing, with ML-algorithms such as RF and GB to model both long-term trends and nonlinear-fluctuations in waste-generation. Secondary data for the period 2015‐2024 were compiled from municipal-corporation records, CPCB-publications, and national urban development reports. Seasonal-indicators, including festival-periods and monsoon-months, along with climatic variables and selected sociodemographic drivers, were incorporated to enhance model responsiveness. Forecast performance was evaluated using RMSE, MAPE, R 2 , and k-fold cross-validation. Results indicate that standalone ARIMA and Holt-Winters models achieved average R 2 values of 0.86 and 0.88, respectively, while ML-models improved predictive accuracy beyond 0.91. The proposed hybrid ensemble consistently outperformed both statistical and ML approaches, recording R 2 values above 0.94 and achieving approximately 20‐25% reductions in RMSE and MAPE, while preserving interpretability for policy-oriented applications. Seasonal decomposition and feature-importance analysis further reveal that festival periods contribute to short-term MSW increases of about 8‐12%, and monsoon-induced moisture variations significantly influence daily waste variability and collection efficiency.