Application of K-Means and K-Medoids Algorithms for Clustering Chili Commodity Trade Distribution in Indonesia

Febri Widianto; Elika Thea Kirana; Nurahman; Depi Rusda

doi:10.20527/jtiulm.v9i2.220

Febri Widianto Universitas Darwan Ali
Elika Thea Kirana Universitas Darwan Ali
Nurahman Universitas Darwan Ali
Depi Rusda Universitas Darwan Ali

DOI: https://doi.org/10.20527/jtiulm.v9i2.220

Keywords: BPS, Clustering, Chili, K-Means, K-Medoids

Abstract

Chili is one of the important commodities in agriculture and food, which is a product of the capsicum plant that has significant economic value in international trade. This study aims to identify an effective distribution strategy for red chili commodities in Indonesia through the use of the K-means and K-medoids clustering algorithms. The data used comes from the Central Statistics Agency (BPS) in 2022, including parameters-production, consumption, surplus/deficit, trade margin, and the impact of market operations and natural disasters. The implementation of K-means and K-medoids uses the RapidMiner application to form six provincial clusters based on the characteristics of red chili distribution. The results of the analysis show that K-medoids consistently outperforms K-means in cluster formation, with lower Davies-Bouldin Index (DBI) values indicating better clusters. The conclusion of this study confirms that K-medoids is more effective in grouping red chili distribution areas in Indonesia, potentially providing a stronger foundation for strategic decision making in the distribution management of this commodity. Therefore, this study recommends the use of K-medoids as a more appropriate approach for planning and implementing red chili distribution strategies in Indonesia.