Clustering of Indonesian Provinces based on Demographic Characteristics using DBSCAN Algorithm

Abstract

This research aims to implement the Density-Based Spatial Clustering of Applications with Noise (DBSCAN) algorithm for clustering provinces in Indonesia based on demographic characteristics using a Web GIS system. The study utilizes demographic data of 34 provinces from 2020–2023, covering population size, population growth rate, poverty percentage, population density, and gender ratio. DBSCAN was applied with parameters epsilon (ε) = 0.50 and minimum samples = 2, evaluated using the Davies-Bouldin Index (DBI). The results show that DBSCAN successfully identified two consistent main clusters throughout the period: the High Cluster consisting of three provinces (West Java, Central Java, East Java), the Medium Cluster including 29–30 provinces, and DKI Jakarta as a single noise point.. Clustering quality varied with DBI values ranging from 0.3839 to 0.4123, with the best quality in 2020 and the highest internal diversification in 2023. Riau Islands fluctuated as a noise point in 2021 before returning to stability. The developed Web GIS system successfully integrated interactive map visualization using OpenStreetMap and Leaflet, providing a comprehensive spatio-temporal analysis dashboard. This study demonstrates the effectiveness of DBSCAN in detecting demographic patterns and regional anomalies, while producing a visualization tool that supports regional development planning and national demographic policies in Indonesia.