Assessing the determinants of municipal solid waste generation in Europe: A machine learning approach.

Waste management is significant for achieving climate goals, considering its contribution to global greenhouse gas emissions. While differences in waste management systems and generation rates across European countries are closely linked to socio-economic, demographic, and environmental factors, previous studies have predominantly focused on specific regions or local municipalities. This study addresses this gap by analyzing Europe as a single entity, providing an overview of the current waste management practices and legislation, assessing influencing factors and employing machine learning (ML) models for MSW generation rates estimation. A comprehensive dataset includes 14 socio-economic, demographic and environmental factors collected for 40 European countries over 2000-2021. To capture regional differences more effectively, the countries were further grouped, and separate models were developed in addition to the general model. Three ML models, namely Artificial Neural Network (ANN), Support Vector Regression (SVR), and Kernel Ridge Regression (KRR) were developed and evaluated. SHAP (SHapley Additive exPlanations) analysis was applied to assess the contribution of individual factors to the predictions of MSW generation rates. Despite being trained on heterogeneous data, all models demonstrated high predictive ability. In all developed models, KRR demonstrated the best performances for the general model and for models for high, medium and low-income countries. The most significant variable in estimating MSW generation rates is average annual wages. The findings of this study may help in bridging the gap between EU and non-EU countries, emphasizing the necessity for robust environmental policies and cross-national collaboration.