Analisis Perbandingan Algoritma K-Means dan K-Medoids dalam Penentuan Status Gizi Balita

Abstract

Nutritional status in toddlers is an important indicator in determining child growth and development quality. Inaccurate classification of nutritional status can affect early intervention efforts. This study aims to compare the performance of K-Means and K-Medoids algorithms in clustering toddler nutritional status data at Puskesmas Betun. The dataset consists of 1,036 toddler records with variables including age, weight, height, and mid-upper arm circumference (MUAC). Data preprocessing was conducted through normalization before clustering. The performance of both algorithms was evaluated using the Davies Bouldin Index (DBI). The results show that K-Means converged in 24 iterations with a DBI value of 1.0281, while K-Medoids converged in 6 iterations with a DBI value of 1.1236. Based on the DBI evaluation, K-Means produced better clustering performance compared to K-Medoids. Therefore, K-Means is more suitable for determining toddler nutritional status in this study.