Clustering Analysis Of Districts/Cities In Sumatra Region Based On Poverty Percentage Using The K-Medoids Method

Abstract

Clustering is a data grouping method applied to identifies groups formed by combining elements that have the same characteristics. One of the clustering methods that can be used is the K-Medoids method known as Partitioning Around Medoids (PAM). This study aims to obtain grouping and determine the characteristics of the results of grouping regencies/cities in the Sumatra Region based on the percentage of poverty using the K-medoids cluster method. The data used are poverty data per district/city totaling 154 in the Sumatra Region with the variables used being the expected length of schooling, average length of schooling, open unemployment rate, and percentage of poor population. The results obtained in this study are that districts/cities in the Sumatra Region have 2 optimum clusters as seen from the silhouette index value and davies-bouldin index value