Klasterisasi Data Untuk Sistem Rekomendasi Pemilihan Sekolah Dasar Di Wilayah Surakarta Degan Metode Hierarchy Clustering

Abstract

The role of parents is very important in determining the school for their children, especially at elementary school level. There are many factors that parents consider when choosing an elementary school, including school quality, distance to school, school fees, content of religious education, school performance and school facilities. The aim of this research is to develop a data clustering concept using Hierarchy Clustering method for an elementary school recommendation system in the city of Surakarta, so that it can help parents in determining the elementary school for their children. In this study, agglomerative clustering with Ward linkage was used, this is because among other linkage methods (Single linkage, ward linkage, flexible linkage and complete linkage), Ward linkage has the highest Silhouette index value, namely 0.666. Cluster optimization in this research was determined by looking at the Silhouette index graph by comparing the Silhouette cluster values 1 to 10 for each linkage method, and obtained the cluster optimization value was 4, so the number of groupings used was 4 clusters. The accuracy of the recommendation system is tested using the Mean Average Error (MAE) statistical calculation, with an MAE value of 0.2, which means the recommendation system is suitable for use.