Cluster Analysis on Various Cluster Validity Indexes with Average Linkage Method and Euclidean Distance (Study on Compliant Paying Behavior of Bank X Customers in Indonesia 2021)

Abstract

This study aims to examine the differences in various cluster validity indexes in the grouping of credit customers at Bank X Malang City, Indonesia using the average linkage and Euclidean distance methods. This study uses primary data with the variables used are service quality, environment, mode, willingness to pay, and obedient paying behavior obtained through a questionnaire with a Likert scale through purposive sampling distributed to 100 respondents. The data are then analyzed by clusters using the ward linkage and Euclidean distance methods on various validity cluster indexes, including the Silhouette Index, Krzanowski-Lai, Dunn, Gap, Davies-Bouldin, Index C, Global Sillhouette, Goodman-Kruskal in this study used as a tool analysis. This study uses R software. The results show that the Krzanowski-Lai, Dunn, Gap, Global Sillhouette, and Goodman-Kruskal indices have the same cluster members, as well as the Silhouette and Davies-Bouldin indices. The best cluster indexes are the Silhouette and Davies-Bouldin indexes. All validity indices produce variance between and within the same cluster. The novelty of this study is to compare 8 validity indices, namely Sillhouette Index, Krzanowski-Lai, Dunn, Gap, Davies-Bouldin, Index C, Global Sillhouette, and Goodman-Kruskal simultaneously.