Perbandingan Metode Perhitungan Jarak pada Nilai Centroid dan Pengelompokan Data Menggunakan K-Means Clustering

Abstract

This study will observe the process of grouping data or forming clusters using K-Means clusters with three methods of measuring distances, namely Euclidean distance, Manhattan distance, and Minkowski distance. Observations are more focused on changing the centroid value and the results of grouping data, as well as the number of iterations required. Experimental data amounted to 20, 30, 40, and 50 pieces of data which were grouped into 2 groups. This research also summarizes the application of K-Means clusters which have been widely used in various fields, including Health, Education, and Disaster. The results of grouping data with the three distance measurement methods are not too much different, namely the highest difference is 2 members of the data on 50 test data. The most iterations on 40 test data use the Euclidean distance, namely 7 iterations, and the least iteration on 20 test data uses Minkowski distance i.e. 3 iterations. On the 50 test data it takes 4 iterations. The amount of test data is not directly proportional to the number of iterations needed to reach the cluster in a stable state.