Perbandingan K-Means dan Agglomerative Hierarchical Clustering dalam Pengelompokan Provinsi di Indonesia Berdasarkan Tingkat Kepemilikan JKN

Abstract

The National Health Insurance (JKN) is a government program designed to achieve Universal Health Coverage; however, its ownership rate varies across provinces in Indonesia. This study aims to cluster the 38 provinces based on JKN ownership levels during 2023–2025 and compare the performance of the K-Means and Agglomerative Hierarchical Clustering (AHC) methods using the Ward Linkage approach. Data were obtained from Statistics Indonesia (BPS) and processed using Python in Google Colab through preprocessing, optimal cluster determination, clustering, and evaluation using the Silhouette Coefficient and Davies-Bouldin Index (DBI). The results indicate that the optimal number of clusters is three, representing low, medium, and high categories. K-Means achieved a Silhouette Coefficient of 0.595773 and a DBI of 0.453632, while AHC obtained a Silhouette Coefficient of 0.587456 and a DBI of 0.433335. Based on the higher Silhouette Coefficient, K-Means is recommended as the more effective method to support policy evaluation and the equitable distribution of JKN participation across Indonesia.