Improving circularity in municipal solid waste management through machine learning in Latin America and the Caribbean

Machine Learning (ML) consists of a set of methods that allow a system to learn data patterns and has applications in many stages of MSWM. Improvements in MSWM focused on resources recovery in LA&C can be speed up by the use of the ML methods. This study aims to analyze the opportunities and challenges faced when using the ML methods to improve circularity in MSWM in LA&C. The methodology adopted was a systematic literature review using the PRISMA protocol in the Web of Science (R) database from 2010 to 2021, and bibliometric analysis using the Biblioshiny (R) application, the web interface for Bibiometrix (R) package from Rstudio (R) software. A total of 188 papers were obtained from the bibliographic search. The advancement of MSWM in LA&C has as challenges the lack of reliable data on the composition and production of the waste, the low rate of waste used as a resource, the need to change consumption patterns, social inclusion of informal waste collectors, and the inclusion of repair and reuse actions to reduce waste generation. Meanwhile, the main challenges when considering the use of ML in LA&C are the inexistence or dispersion of data with reliable time series and the lack of knowledge of decision makers about the potential use of the ML methods. Specifically in LA&C, it was observed that hybrid models that apply ML to waste composition data, ML methods to improve IoT applications and GIS data usage aggregated with ML methods could speed up the transition to Circular Economy in LA&C.

Automated summary

This study analyzes the opportunities and challenges faced when using machine learning methods to improve circularity in management of solid waste in Latin America and the Caribbean (LA&C). The authors conducted a systematic literature review and bibliometric analysis and found that the challenges in improving solid waste management in LA&C include lack of reliable data, low waste utilization rate, need for changing consumption patterns, social inclusion of informal waste collectors, and limited knowledge of decision makers about the potential of machine learning methods. Hybrid models, improved Internet-of-Things (IoT) applications, and integration of Geographic Information System (GIS) data with machine learning methods are identified as potential ways to accelerate the transition to circular economy in LA&C.