Application of Ensemble K-Modes and SWFM for Grouping Sulawesi Tengah Regions by Underdeveloped Indicators

Abstract

This research aims to determine the best final clustering results and clustering statistics for regencies/cities in Central Sulawesi based on underdeveloped region indicators. The study uses categorical and numerical data variables, consisting of 10 numerical variables and 3 categorical variables. The methods used in this research are the mixed data Ensemble K-Modes and the Similarity Weight and Filter Method (SWFM). The best mixed data clustering method shows that the Ensemble K-Modes method produces better clustering results than the SWFM method, as Ensemble K-Modes has a higher accuracy score of 0,8462