ANALISIS k-MEDOIDS DENGAN VALIDASI INDEKS PADA IPM DAERAH 3T DI INDONESIA

Abstract

Human development is a development paradigm that places humans as the main target of all development activities, namely controling over resources, improving health and improving education. The Human Development Index (HDI) in Indonesia varies in each district, especially in the 3T areas. The 3T area is an area that is classified as underdeveloped, remote and outermost in terms of economy, health, education and infrastructure. The k-Medoids method is a partitional clustering method for grouping several objects into clusters. This clustering algorithm uses the medoid as the center of the cluster, so it is robust to data containing outliers. This study aims to classify the 3T regions in Indonesia based on the Human Development Index to find out which areas require more attention from the government in optimizing the Human Development Index numbers. The size of object similarity is calculated by using the Euclidean distance and Manhattan distance, for the selection of the best number of clusters, internal cluster validation, such as Calinski – Harabasz index, Gamma Index, and Silhouette index. The result of this study showed that the best cluster were four by using Euclidean distance measurement, having Calinski – Harabasz index score of 37.15764, Gamma index score of 0.7821181, and Silhouette index score of 0.3354435.