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

Authors

Adji Achmad Rinaldo Fernandes, Department of Statistics, Brawijaya University, Malang, IndonesiaFollow
Solimun Solimun, Department of Statistics, Brawijaya University, Malang, IndonesiaFollow
Erlinda Citra Lucki Efendi, Department of Statistics, Brawijaya University, Malang, IndonesiaFollow
Ni Made Ayu Astari Badung, Department of Statistics, Brawijaya University, Malang, IndonesiaFollow
Endang Krisnawati, Department of Statistics, Brawijaya University, Malang, IndonesiaFollow

Abstract

This study aims to examine and explain the differences in the use of various cluster validity indices in the application of KPR customer grouping at Bank X Malang City with various linkages and euclidean distances. The data used in this study are primary data. The variables used in this study are as follows: quality of service, environment, fashion, willingness to pay, and obedient behavior to pay at Bank X. The object of observation is the customer as many as 100 respondents using purposive sampling technique. Data analysis was carried out quantitatively using cluster analysis with 4 cluster validity indices, namely the Silhouette, Krzanowski-Lai, Dunn, and Davies- Bouldin indexes. The results showed that the difference in the validity index did not affect the number of clusters produced. In addition, differences in the linkage method used affect the number of clusters generated as well as the variance within and between clusters. This study also shows that the best method to examine and explain the differences in the use of various cluster validity indices in the application of grouping mortgage customers at Bank X Malang City is the hierarchical clustering technique with the average linkage method with the Davies-Bouldin index value ranging from 0.47 to 0.55. The novelty in this study is to compare 4 validity indices, namely the Silhouette Index, Krzanowski-Lai, Dunn, Davies-Bouldin on various linkage methods, including ward, average, and complete linkage simultaneously.

Digital Object Identifier (DOI)

http://dx.doi.org/10.18576/jsap/110311

Recommended Citation

Achmad Rinaldo Fernandes, Adji; Solimun, Solimun; Citra Lucki Efendi, Erlinda; Made Ayu Astari Badung, Ni; and Krisnawati, Endang (2022) "Cluster Analysis Study on Various Cluster Validity Indexes with Various Linkages and Euclidean Distance (Study on Compliant Paying Behavior of Bank X Customers in Indonesia 2021)," Journal of Statistics Applications & Probability: Vol. 11: Iss. 3, Article 11.
DOI: http://dx.doi.org/10.18576/jsap/110311
Available at: https://digitalcommons.aaru.edu.jo/jsap/vol11/iss3/11