APPLICATION OF DBSCAN FOR CLUSTERING SOCIETY BASED ON WASTE MANAGEMENT BEHAVIOR

Abstract

This research aims to answer the challenge of identifying the characteristics of the Batu City community in waste management, where traditional clustering techniques are often suboptimal due to the presence of noise or objects that do not fit the general pattern. As a solution, the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm is applied, which allows the clustering of objects based on local density and detects the presence of noise or outliers in the data. DBSCAN is considered more flexible than other clustering methods, especially in clustering data that is not linear or has a non-uniform distribution. This study successfully identified three clusters of waste management behavior with a silhouette index of 0.875, indicating good cluster quality. The first cluster consists of communities with good environmental quality, active participation in the use of waste banks, and a deep understanding of 3R-based waste management. The second cluster has adequate infrastructure quality and high awareness of the potential economic benefits of waste, while the third cluster displays a pretty good level of understanding of the 3Rs and relatively good environmental quality. The results of this study provide important insights into the differences in waste management characteristics between clusters, with environmental quality proving to be a significant factor in cluster formation.