Comparison of Non-Hierarchical Clustering Performance on the Regional Competitiveness Index in Central Java in 2022

Abstract

Regional Competitiveness Index (RCI) is a benchmark to measure the ability of a region to compete in a market published by the National Research & Innovation Agency (BRIN). RCI includes several pillars or indicators including infrastructure, quality human resources, innovation, and government policies that support economic growth. This study aims to compare the performance of several non-hierarchical clustering techniques. The data used are the RC) from 35 Regencies/Municipalities in Central Java,2022 which was published by the National Research and Innovation Agency. The clustering methods used are Fuzzy c-means, K-means, and K-medoid. Each method gets a different optimal number of clusters. After evaluating the best model using the Silhouette Coefficient, Dunn Index, Davies Bouldin Index, Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC), the best model was obtained using k-medoid with three clusters. Based on the clusters formed, the first cluster has three regencies/municipalities, the second cluster has regencies/cities, and the third cluster has 25 regencies/ municipalities.