PENINGKATAN KINERJA ALGORITMA K MEANS DENGAN MENGGUNAKAN PARTICLE SWARM OPTIMIZATION DALAM PENGELOMPOKAN DATA PENYEDIAAN AKSES

Abstract

Water is one of the things that plays a very important role in human survival, because the Indonesian government has a community-based water supply and sanitation (PAMSIMAS) program, so that all the programs run well need a regional status grouping technique in this thesis. with the K-means algorithm.
 K-means is a partition algorithm that aims to divide the data into the specified number of clusters, the results of the K means algorithm depend on the selection of the initial klater center but problems that often occur when selecting the initial centroid are randomly drawn from the solution. from the grouping is not quite right. To overcome this problem the author wants to use the PSO algorithm in the initial centroid selector for the K-means algorithm, in this study also compared the selection of the first 3 centroids according to random, second according to government standards the value of high, medium and low drinking water quality then the third method proposed by the PSO algorithm was then tested with Davies Bouldin Index.
 From the test results, the K-means method with the selection of random initial centroid with a value of 0.208856082, the K-means method with the selection of centroids in accordance with government standards about SAM conditions of 0.280077 and the best selection method is K-means PSO 0, 08383. So testing the PAMSIMAS data using K-means PSO found that the method was more optimal.