Clustering Algorithm-Based Pandemic Multicluster Framework Analysis: K-Means and K-Medoids

Several research results show that the variables determining the distribution of the pandemic become a reference for all regions in dealing with the pandemic. A number of frameworks were built to assist in determining the dominant variables causing the pandemic, including a framework with a multi-cluster approach that was formed based on assumptions. This study aims to explore the ability of the multi-cluster framework to determine the dominant variables causing the pandemic through the Davies Bouldin Index (DBI) cluster evaluation technique with the K-Means and K-Medoids algorithms. In the first assumption, the test results show that there are similarities in cluster members for both algorithms. The second assumption produces different cluster members from the two algorithms, where the optimal K-Means is at K=5 with the predominance of male gender criteria, the eldest age group, high school graduated and employee occupation. For K-Medoids, the optimal cluster is at K = 8 with the predominance of criteria for female gender, adult age group, high school graduated and job not working.