https://doi.org/10.30598/barekengvol18iss2pp1167-1178 June 2024 Volume 18 Issue 2 Page 1167–1178 P-ISSN: 1978-7227 E-ISSN: 2615-3017 BAREKENG: Journal of Mathematics and Its Applications 1167 PROVINCIAL CLUSTERING BASED ON EDUCATION INDICATORS: K-MEDOIDS APPLICATION AND K-MEDOIDS OUTLIER HANDLING Octavia Rahmawati 1 , Achmad Fauzan 2* 1,2 Statistics Study Program, Faculty of Mathematics and Natural Sciences, Universitas Islam Indonesia Jl. Kaliurang KM 14.5, Sleman, 55584, Indonesia Corresponding author’s e-mail: * achmadfauzan@uii.ac.id ABSTRACT Article History: K-Medoids is a clustering algorithm that is often used because of its robustness against outliers. In this research, the focus is to cluster provinces based on educational level through several assessment indicators. This is in line with improving the quality of education in point 4 of the National Sustainable Development Goals (SDGs), namely "Quality Education". One of the points of the National Sustainable Development Goals (SDGs) that will still be improved is "Quality Education" which is the 4th point. This is because the success of a country is determined by the quality of good education. The condition of education in Indonesia still overlaps, so it is necessary to do equal distribution of education through clustering. The purpose of this research is to provide the best cluster results according to the Silhouette Index, which then the results of the clustering can be used as a consideration for advancing education in areas that still need attention, through policies or programs that can be developed by educational observers. This research was conducted in 34 provinces in Indonesia. The data source is from Statistical Publications by BPS RI. The method used is K-Medoids, because in this study there were outliers found. In addition to natural K-Medoids, the researcher also wants to compare methods by implementing K-Medoids with outlier handling in the form of imputed mean values and K-Medoids with imputed min-max values. The Silhouette Index results and cluster formation for the three comparators were 0.24 with 2 clusters, 0.26 with 8 clusters and 0.25 with 9 clusters, respectively. This research differs from previous work in its approach to outlier handling. While K-Medoids is a straightforward clustering method and generally indifferent to outliers, its effectiveness can be reduced by local outliers and random initial medoid selection. Received: 3 rd January 2024 Revised: 6 th February 2024 Accepted: 28 th March 2024 Published: 1 st June 2024 Keywords: Education; K-Medoids; Outlier Imputation. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution-ShareAlike 4.0 International License. How to cite this article: O. Rahmawati and A. Fauzan., “PROVINCIAL CLUSTERING BASED ON EDUCATION INDICATORS: K-MEDOIDS APPLICATION AND K-MEDOIDS OUTLIER HANDLING,” BAREKENG: J. Math. & App., vol. 18, iss. 2, pp. 1167-1178, June, 2024. Copyright © 2024 Author(s) Journal homepage: https://ojs3.unpatti.ac.id/index.php/barekeng/ Journal e-mail: barekeng.math@yahoo.com; barekeng.journal@mail.unpatti.ac.id Research Article ∙ Open Access