ANALISIS CLUSTER DENGAN ALGORITMA K-MEANS DAN FUZZY C-MEANS CLUSTERING UNTUK PENGELOMPOKAN DATA OBLIGASI KORPORASI

Abstract

Cluster analysis is a method of grouping data (object) that are based on information that found in the data which describes the object and relation within. Cluster analysis aims to make the joined objects in the cluster are identical (or related) with one another and different (not related) to objects in another cluster. In this study used two method of grouping; Fuzzy C-Means and K-Means Clustering. The data used in this research had been using 357 corporate bonds data on December 1 st , 2015. The variables used in this study consist of coupon rate, time to maturity, yield and rating of each corporate. The determination of the number of optimum clusters performed by Xie Beni index of validity calculation at FCM method. Having obtained the optimum number of clusters, evaluation step was conducted by comparing FCM method to K-Means method with noticing the average of standard deviation in the clusters and the average of standard deviation inter-clusters (Sw/Sb) from each method. Method with the smallest Sw/Sb ratio value would get chosen as the best method. Based on the validity index Xie Beni, the most optimum number of cluster is 10 because it has the smallest Sw/Sb ratio value compared to FCM, the value is 0,6651. Afterwards, the result of K-Means clustering is analyzed to determined the interpretation and characteristics of each formed clusters. Keyword: Cluster Analysis, coupon rate, time to maturity, yield, rating, Fuzzy C-Means, K-Means, Xie Beni Index, Sw/Sb ratio.