Municipal Solid Waste Management Using Machine Learning: A Case Study in Sheger City, Koye Sub-city, Ethiopia

Municipal Solid Waste Management is an increasingly critical challenge in urban areas, intensified by rapid urbanization, population growth, and evolving consumption patterns. This study investigates the application of machine learning techniques to predict municipal solid waste generation in Sheger City, Koye Sub-city, Ethiopia, using data from 2009 to 2023. Three machine learning models, ARIMA, RF, and LSTM, were employed to forecast waste generation trends for the period 2024–2028, considering various socio-economic and demographic factors. Among the models, LSTM demonstrated the highest accuracy, with MSE of 1.62 × 10⁸ tonnes, MAE of 9,500 tonnes, and R² of 0.93. These results outperformed ARIMA (MSE = 3.84 × 10⁸ tonnes², MAE = 15,200 tonnes, R² = 0.85) and RF (MSE = 2.91 × 10⁸ tonnes², MAE = 12,800 tonnes, R² = 0.89). The forecasts predict an 8.5% increase in total waste generation, from 3,852,150 tonnes in 2023 to 4,177,500 tonnes by 2028. Notable growth is expected in high-volume waste streams, including food waste (13.5% increase) and plastic waste (8.9% increase). These findings highlight the urgent need for enhanced waste management strategies, including expanded recycling programs and policy interventions. This study provides a robust framework for leveraging machine learning models to guide waste management decisions, contributing to more sustainable urban waste management practices in rapidly growing cities.