Pengelompokan Data Kategorik Dengan Algoritma Robust Clustering Using Links

Studi Kasus: PT. Prudential Life Jalan MT. Haryono Samarinda

  • Isma Dewi Laboratorium Statistika Komputasi FMIPA Universitas Mulawarman
  • Syaripuddin Syaripuddin Laboratorium Statistika Komputasi FMIPA Universitas Mulawarman
  • Memi Nor Hayati Laboratorium Statistika Terapan FMIPA Universitas Mulawarman

Abstract

Cluster analysis is a technique of data mining that is used to group data based on the similarity of attributes of data objects. The problem that is often encountered in cluster analysis is the data on a categorical scale. Categorical scale data grouping can be done using the ROCK (RObust Clustering using linKs) algorithm. The ROCK algorithm is included in the of agglomerative hierarchical clustering algorithms in cluster analysis. This algorithm introduces a concept called neighbors and links in grouping data. Categorical data grouping with ROCK algorithm is done in three steps. The first step is counting similarities. The second step is determining the neighbors and the last is calculating the links between the observation objects. The value of the link is affected by θ. The optimum number of clusters in the ROCK algorithm is selected using a minimum ratio value of . The purpose of this study is to group 100 data of insurance customers of PT. Prudential Life Samarinda in 2018. Based on the analysis results, obtained that the optimum group is at θ = 0.1 with a ratio value of  is 0.1371. The optimum number of groups formed is 2 clusters. The first group consisted of 42 customers and the second group consisted of 58 customers.

Published
2021-01-19
How to Cite
DEWI, Isma; SYARIPUDDIN, Syaripuddin; HAYATI, Memi Nor. Pengelompokan Data Kategorik Dengan Algoritma Robust Clustering Using Links. JURNAL EKSPONENSIAL, [S.l.], v. 11, n. 2, p. 133-138, jan. 2021. ISSN 2085-7829. Available at: <http://jurnal.fmipa.unmul.ac.id/index.php/exponensial/article/view/655>. Date accessed: 23 june 2021.