Perbandingan Metode C-Means dan Fuzzy C-Means Dalam Pengelompokkan Wilayah Desa/Kelurahan di Kabupaten Kutai Kartanegara

Abstract

Cluster analysis is a multivariate statistical technique that has the main purpose to classify objects based on common characteristics. With this analysis, the object will be grouped such that each object is the closest similarity to other objects are in the same group. In the clustering process by using no hierarchical C-Means formation of partition is done such that each object explicitly declared as a member of one group and not a member of any other group. But sometimes can not put an object just in one partition, because in fact the object is located between two or more other partitions, so it needs to be weighted based on its fuzzy membership level. In this way, it is to define a method in the formation of the group will be more flexible. The concept is called fuzzy clustering, the fuzzy way each object can be members of multiple groups. The difference lies in the assumptions used as a basis for allocation. One technique that is not part of the method of using the hierarchical nature of fuzzy clustering technique is using Fuzzy C-Means (FCM). This study will examines comparative method C-Means and FCM clustering in a case study, namely the grouping of the village/urban village in Kutai Kartanegara regency based on the characteristics of facilities/infrastructure and socio-economic factors of the population. The results showed that in some respects, FCM was superior than the C-Means, especially in generating the minimum of objective function, the computation time and ratio value S_w and S_b. Based on the similarity matrix eigen value and the index value Xie and Beni (XB) concluded that the most optimal number of groups is 5 (five) groups.