Perbandingan Hasil Penggerombolan Metode K-Means, Fuzzy K-Means, dan Two Step Cluster

Abstract

The main principle of cluster analysis is to classify objects into clusters based on similarity measures. K-means and fuzzy k-means can be classified as popular clustering methods, which are suitable for large data with continuous variables. However, a new method has been developed to be used for large data, that is the two step cluster method. This method allows processing data with different types of variables, which in this case are continuous and categorical. The aim of this research is to compare the clustering results of k-means, fuzzy kmeans, and two step cluster method, in order to determine the ideal number of clusters for each method. This research uses hypotetical data taken from SPSS software, which fit the purpose to compare several methods. The results of this study show that in the case of two clusters, k-means and fuzzy k-means methods have more similarities with respect to the number objects in clusters, whereas the two step method gives unequal number of objects in clusters. All methods show that 2 clusters is an ideal number. It is influenced by the ratio between mean squares within clusters, which is smaller than the ratio in the case of 3 and 4 clusters.