CLUSTERING OF REGIONS WITH POTENTIAL FOR A TSUNAMI IN INDONESIA USING THE DBSCAN METHOD (DATA STUDY FOR 1822-2022)
Melany Febrina, Avisena

Department of Physics, Institut Teknologi Sumatera

Abstract

Geographically, Indonesia is an archipelagic country that has a long coastline.
Many community activities are carried out in coastal areas, especially coastal
communities. Tsunami is one of the risks of natural disasters that can occur in the area. This study aims to classify areas prone to tsunamis and classify their
characteristics. The variables in this study are longitude, latitude, focal depth, and earthquake magnitude. In this study, Density-Based Spatial Clustering of
Application with Noise (DBSCAN) and OPTICS algorithms were used to group
tsunami datasets. To test the quality of the model, the silhouette score calculation method was used. The results of DBSCAN clustering with =1.9 and MinPts=3 obtained 9 clusters with a silhouette score of 0.378835. Meanwhile, the OPTICS clustering method with =1.9 and MinPts=3 obtained 18 clusters with a silhouette score of 0.242401.

Keywords: DBSCAN, OPTICS, tsunami clustering, MinPts, silhouette score

Topic: Industry 4.0