IJSRP, Volume 10, Issue 7, July 2020 Edition [ISSN 2250-3153]
Adella Sari Cahyani Sugiono, Indahwati, I Made Sumertajaya
Cluster analysis is a multivariate analysis that classifies objects based on their characteristics. Clustering analysis is generally used in cross-section data, that typically taken one point in time and unlike panel data that are taken at multi-times of several objects. This study explores methods for clustering analysis of panel data via distance measures. The objective of this research is to compare the Manhattan Distance, Euclidean Distance, Maximum Distance, Frechet Distance and Dynamic Time Warping (DTW) Distance for Clustering Analysis of Panel Data. The best of the distance measure was implemented empirically for Clustering of Indonesians Province that based on the Human Development Index (HDI), from 2010-2019. Results show that the Manhattan, Euclidean and Maximum provide distances with optimum performances, when the generated data between the clusters are not overlapping. However, when there were overlaps between clusters, the Manhattan distance was the most appropriate.