Pengelompokan Algoritma K-Means dan K-Medoid Berdasarkan Lokasi Daerah Rawan Bencana di Indonesia dengan Optimasi Elbow, DBI, dan Silhouette

Abstract

The study examines the use of K-Means and K-Medoids algorithms in the grouping of disaster area locations in Indonesia, with the aim of identifying patterns and optimizing disaster re-sponse strategies. The data used includes geographical and historical information of various disaster events in Indonesia, such as Aceh Besar, Asahan, Badung, Bangkalan, Bekasi, and others. In the clustering process, optimization techniques such as the Elbow Method, the Da-vies-Bouldin Index (DBI), and the Silhouette Score are used to determine the optimal number of clusters. Research results show that the K-Means algorithm tends to be more stable in deal-ing with outliers than K- Means, with the results of the DBI (Davis-Booldin Index) 0.3737248981 and the cluster 7, resulting in the silhouette score of 0.868728638 and cluster 2, resulting at the elbow 98106477130.371 and claster 2. The Silhouette Score and Elbow index-es also provide a strong indication that the clustering algorithm used is capable of forming significant and meaningful clusters. The study has made important contributions to the opti-mization of clustering with three methods used so that it can be the basis for authorities in planning and implementing more effective disaster mitigation policies.