Determine the optimal number of clusters kmeans
1Handry Eldo, Syahril Efendi, Herman Mawengkang
K-means clustering technique has been very widely used in various fields such as academics, practitioners and so on. However, k-means itself still has some shortcomings, including the problem of the accuracy of the algorithm used to measure the similarity between objects being compared. To overcome this problem, the optimum number of clusters will be calculated in this study (euclidean distance, Manhattan distance, and Chebyshev distance) to find out the optimum number of clusters. Silhouette Coefficient test results for each distance measure, including Euclidean Distance worth 0.232149, Manhattan Distance worth 0.240016, and Chebyshev Distance worth 0.242821. Based on the results of the silhouette coefficient testing conducted, the most optimal distance measure for this case is Chebyshev Distance, that is, the silhouette coefficient value closest to 1 is 0.242821.
k-means, cluster, Euclidean, Manhattan