Cluster Application with K-Means Algorithm on the Population of Trade and Accommodation Facilities in Indonesia

Abstract

The aim of this study is to develop a grouping model in order to determine the means of trade and accommodation according to the regions in Indonesia. Research can be a reference for the government to increasing the income of each region in Indonesia equally. Research data were taken from a website that provides government statistical data, namely BPS (Badan Pusat Statistik)-www.bps.go.id The solution is to use data mining techniques with clustering methods. The data test process uses the Rapid Miner software. Three clusters of mapping labels are used, namely the high cluster (K1), the normal cluster (K2) and the low cluster (K3). The results of the rapidminer processing were obtained from the centroid data for high clusters, namely ((1527), (810.4), (5865), (6655.3), (323), (315.1); the medium cluster, namely ((286), (199.591), (1327), (2240.227), (93.227), (140.955)); and the low cluster, namely (139.25), (122.5), (508.833), (919.222), (64.417), (94.444)). The cluster results show that 5 provinces are classified as high in clusters; 13 provinces are classified as medium clusters; and 16 provinces are classified as low clusters. Out of the results of the study, some 47% of areas in Indonesia still have low trade and housing facilities. With this analysis, it is hoped that the government will be able to pay more attention to regions whose revenues are still below average.