Deteksi Pencilan pada Data Titik Panas Menggunakan Clustering Berbasis Medoids

Abstract

Forest fire is one of disasters which has a very adverse impact. Land and forest fires in Indonesia are caused by several factors, such as prolonged drought, human negligence and irresponsible parties who deliberately set fire to achieve certain goals. Hotspot is an indicator of forest fires. The purpose of this study is to detect outliers in hotspots in 2001 until 2012. Hotspot data were obtained from the NASA FIRM. The outlier detection was performed using medoid based clustering methods, namely PAM and CLARA. The result of PAM algorithm show that outliers occur in cluster k=17 with medoid 13,14,15,16 and 17. The result of CLARA algorithm show that outliers occur in cluster k = 19 with medoid 14,15,17 and 19. PAM and CLARA algorithm detect outliers in February, March, June, July and August. Clustering results are expected to assist the authorities in determining potential areas for forest fires prevention