Perbandingan Metode K-Means, K-Medoids, dan Hierarki untuk Penggerombolan Kabupaten/Kota di Sumatera Utara Berdasarkan Jenis Kekerasan terhadap Perempuan

Abstract

Violence is an act that is intentionally by a person or group with the aim of oppressing someone to suffer. The high rate of depression which leads to acts of violence against those around them, including women. This study aims to compare the clustering of districts/cities in North Sumatra based on the type of violence against women using the K-Means, K-Medoids, and Hierarchical Methods, and look at the characteristics of the regional distribution of the clustering results. This study uses data the number of victims of violence against women in 2021 from Sistem Informasi Online Perlindungan Perempuan dan Anak and Badan Pusat Statistik Sumatera which consists of 33 districts/cities and six variables. Evaluation of clustering results using the Davies Bouldin Index and Cophenetic Correlation. The optimal cluster analysis method obtained is K-Medoids with 4 clusters. Cluster 1 has the lowest average value for each type of violence. Group 2 has an average value of physical violence that is higher than the average. Group 3 has a higher average score for the types of psychological violence, trafficking, and neglect than the average. Cluster 4 had a highest average score for each type of violence except trafficking than the average.