Implementasi Metode K-Means Untuk Clustering Data Penduduk Miskin Dengan Systematic Random Sampling

Abstract

The poor population grouping aims to differentiate the population with the highest or the most appropriate level of poverty to get assistance specifically for the population with the highest level of poverty. Grouping is done by using the k-means method. Grouping with the k-means method will increase the level of similarity in groups and reduce the level of similarity between groups. Random grouping on k-means will be applied systematic random sampling methods that will influence and narrow down the possibility of many initial centroid values ??to be generated, while speeding up the computation process for random grouping. Furthermore, the silhouette coefficient is validated to determine the best group in grouping the poor population. The number of groups determined is 2 clusters, 3 clusters, and 4 clusters. The results obtained are the number of groups of 2 clusters is better than 3 clusters and 4 clusters with a value of 2 clusters namely 0.435489, while in 3 clusters 0.434857 and 4 clusters 0.30832.