Clustering Locations of Collection Centers in Reverse Logistics Network for Municipal Waste Management Using Machine Learning

Waste generation is a significant issue in sustainable development, driven by rapid population growth and urbanization. In Indonesia, waste generated from 177 districts/cities in 2023 will reach 20,589,790.62 tons/year. Waste management aims to reduce environmental impacts and human health by collecting, transporting, processing, and disposing of waste. Reverse logistics (RL) is a green supply chain management practice that reduces waste from end-of-life products or packaging waste. Surabaya as a metropolitan city in Indonesia, faces increasing population density and increased waste generation. To optimize waste management, this research aims to cluster waste sources in Surabaya using an unsupervised machine learning method. The study found that cluster 1 is the most suitable cluster with the most waste generated (1,936,016.2 kilogram/year), while cluster 2 is the least generated (107,658.5 kilogram/year). However, the study has limitations in comparing other clustering techniques, and future research should develop other algorithms to determine clustering locations in collection centers and integrate two methods, clustering and optimization, to evaluate the implications of clustering waste sources for optimizing the reverse logistics network.