ASSESSING CLUSTER VALIDITY AND STABILITY OF HIERARCHICAL WARD’S LINKAGE AND NON-HIERARCHICAL K-MEANS ON THE EBGS INDEX OF REGENCIES/MUNICIPALITIES IN SOUTH SULAWESI

Elisabeth Evelin Karuna (Public Administration Study Program, Faculty of Social Sciences and Law, University of Makassar, A.P. Pettarani Street, Makassar City, 90222, South Sulawesi)
Mahrani Mahrani (Public Administration Study Program, Faculty of Social Sciences and Law, University of Makassar, A.P. Pettarani Street, Makassar City, 90222, South Sulawesi)
Atiqa Azza El Darman (Public Administration Study Program, Faculty of Social Sciences and Law, University of Makassar, A.P. Pettarani Street, Makassar City, 90222, South Sulawesi)

Jurnal Statistika Universitas Muhammadiyah Semarang
Vol 14, No 1 (2026): Jurnal Statistika Universitas Muhammadiyah Semarang

Abstract: Cluster analysis is a statistical method used to group objects based on similar characteristics. In general, there are two main categories in cluster analysis, namely hierarchical methods (such as Ward's linkage) and non-hierarchical methods (such as K-Means). This study aims to compare the performance of these two methods in grouping the Electronic-Based Government System (EBGS) Index of districts/cities in South Sulawesi Province. The results of the analysis show that both methods produce identical validity index values, namely a Silhouette Coefficient of 0.67, a Davies-Bouldin Index (DBI) of 0.39, and a Calinski-Harabasz Index (CHI) of 83.02. These values indicate that the clusters formed have high internal compactness and clear separation between clusters. In addition, the Adjusted Rand Index (ARI) value of 1.00 indicates perfect agreement between the results of Ward's linkage and K-Means, signifying a very high level of stability. Thus, the results of this study show that the grouping of the SPBE Index in South Sulawesi is valid, stable, and able to represent the natural structure of the data consistently. Keywords:Cluster; K-Means; Ward's Linkage; Validity; Stability; EBGS