Implementasi Implementasi K-Means dan K-Nearest Neighbor untuk Klasifikasi Status Gizi Balita (Studi Kasus: Gizi Balita Desa Suka Damai 2024)

Abstract

K-Means and K-Nearest Neighbor are quantitative methods used in data analysis for clustering and classification. K-Means is used to group data based on proximity to the centroid, while K-Nearest Neighbor is used to classify data based on the nearest neighbors. This study aims to analyze the nutritional status of toddlers in Suka Damai Village in 2024 based on growth indicators. The data used consisted of 149 toddlers, including age, weight, height, and head circumference. The number of clusters was determined as five categories according to World Health Organization standards, namely severe malnutrition, undernutrition, normal nutrition, overnutrition, and obesity. The clustering results using K-Means showed that 23 toddlers were categorized as severely malnourished, 29 as undernourished, 49 as normal, 35 as overnourished, and 13 as obese. Cluster evaluation using the Silhouette Coefficient produced an average value of 0.445, indicating that the cluster quality was still relatively weak. Furthermore, classification using the K-Nearest Neighbor method was carried out by dividing the data into 80% training data (119 data points) and 20% testing data (30 data points). The confusion matrix evaluation results showed an accuracy of 100%, with Precision, Recall, and F1-Score values each reaching 1. These results indicate that the K-Nearest Neighbor method has excellent performance in classifying toddlers’ nutritional status, while K-Means is capable of providing an overview of nutritional status distribution although the cluster quality is not yet optimal.

Keywords: Clustering, Classification, K-Means, K-Nearest Neighbor, Toddler Nutritional Status.