Spatio-temporal Clustering Hotspot di Sumatera Selatan Tahun 2002-2003 Menggunakan Algoritme ST-DBSCAN dan Bahasa Pemrograman Python

Abstract

These last few years the issues of forest fire in Sumatera increase dramatically, attracting international attention. Hotspot data which are monitored using satellites can be used as an indicator of the occurrence of fire on the Earth's surface. One approach to analyze hotspot dataset is spatio-temporal clustering which can recognize patterns of hotspot event based on space and time. This study applies ST-DBSCAN clustering using hotspot data in South Sumatera 2002-2003. As multi-paradigm programming language, Python is chosen so that STDBSCAN algorithm can work fast. By using spatial distance (Eps1) 22 km, temporal distance (Eps2) 7 days, and density of cluster (MinPts) 7, 41 clusters were found to have many stationary patterns in Musi Banyu Asin. The average ST-DBSCAN execution runtime using Python was 4.934 seconds