Comparison of K-Means and K-Medoids Algorithms for Clustering of Potential Flood-Prone Areas in Bengkulu Province

This title has been presented on Thursday, December 14, 2023 at 14.10-14.20 GMT+7.

Abstract

This title has been presented on Thursday, December 14, 2023 at 14.10-14.20 GMT+7.

This research examines the potential for flooding in Bengkulu Province, which is a flood-prone region. Factors such as high rainfall, hilly topography, land use change, inadequate drainage systems, and climate change are the main causes of flooding in this region. This research uses the variables of rainfall and river distance to the sub-district center point processed using QGis software. In this study, clustering methods, specifically K-Means and K-Medoids, were used to analyze potential flood-prone areas. The results showed that K-Means with 2 clusters provided better results based on the evaluation of the silhouette coefficient, David Bouldin Index, and Calinski-Harabasz Index. In both methods, Cluster 1 has higher rainfall and the river is further away from the sub-district center than Cluster 2. The development of a Geographic Information System (GIS)-based system can improve understanding and effective management of flood risk by mapping flood-prone areas.