Clustering of Child Stunting Data in Tangerang Regency Using Comparison of K-Means, Hierarchical Clustering and DBSCAN Methods

Abstract

This study aims to analyze stunting in children in Tangerang Regency using clustering methods such as k-means, Hierarchical Clustering with Agglomerative Nesting, and Density-Based Spatial Clustering of Applications with Noise (DBSCAN). Stunting is a significant health issue affecting child growth due to chronic malnutrition and recurrent infections. The research revealed that k-means produced the best clustering results with a Silhouette Score of 0.52, indicating its effectiveness in categorizing children based on age, nutritional status, and stunting risk. The k-means method identified three clusters: Cluster 0 (ages 46-55 months, good nutrition, no stunting), Cluster 1 (ages 9-18 months, varied nutritional status, high stunting risk), and Cluster 2 (ages 27-36 months, good nutrition, no stunting). The study suggests preventive actions such as balanced nutrition education, regular health monitoring, complete immunizations, and physical activity, alongside curative measures like nutritional consultations and supplements. The findings provide a framework for targeted preventive and curative interventions, enabling Tangerang Regency's health department to effectively address and reduce stunting rates.