Insights into the urban municipal solid waste generation during the COVID-19 pandemic from machine learning analysis

Abstract

Municipal solid waste (MSW) management has become an important issue for cities around the world. Waste generation prediction plays an important role in MSW management. It can provide strong support for decision-makers to optimize collection and treatment plans, formulate policies, and design long-term strategies for waste reduction. This study develops a district-level predictive approach for MSW generation using machine learning, with a case study for New York City. Commencing in 2020, the year of the pandemic's outbreak with an abnormal fluctuation in MSW generation, the prediction model exhibited a notable accuracy reduction compared to preceding years. In response to this challenge, we have identified and examined four proxy variables pertinent to the socio-economic repercussions of the pandemic and discerned that variables relating to residents' working status can ameliorate the model performance, both before and after the pandemic. We find that the area with the largest abnormality in MSW generation in 2020 is Manhattan and its vicinity. Given the pronounced prevalence of commercial land use in these localities, they warrant heightened scrutiny and strategic focus. It is imperative for policymakers and stakeholders to channel their efforts towards these specific regions and proactively formulate contingency strategies to facilitate expedited and efficacious responses.