Clustering Data Indeks Pembangunan Manusia (IPM) Pulau Jawa Menggunakan Algoritme ST-DBSCAN dan Bahasa Pemrograman R

Abstract

Human development index (HDI) is an important indicator for regional quality evaluation. HDI data consist of a polygon-shaped spatial and temporal data. ST-DBSCAN can process both spatial and temporal data. The purpose of this research is to apply ST-DBSCAN algorithm on HDI data on Java. Nine clusters are found on Eps1 = 0.4, Eps2 = 0, MinPts = 4 and delta-e equal to 2 local z-score. The highest HDI value is located at Jabodetabek, 77.67. It is caused by its location at capital of Indonesia. This is in contrast with the eastern of Java that has the lowest HDI value, 72.24. The precision of a cluster is calculated using silhouette index. In this research, the cluster evaluation is 0.1599514. Based on this result, it is still not close to the best value. ST-DBSCAN algorithm is implemented using R language, using maptools, cluster and rgeos packages.