IDENTIFYING THE CLUSTER OF FAMILIES AT RISK OF STUNTING IN YOGYAKARTA USING HIERARCHICAL AND NON-HIERARCHICAL APPROACH

Abstract

Stunting, or short stature, is a growth disorder usually caused by chronic dietary deficiencies from the prenatal stage to early childhood, typically becoming evident in children after the age of 2. Stunting cases in Yogyakarta Province experienced a decline in 2020. With this development, the government aims to achieve zero stunting in Yogyakarta Province by 2024. To support this goal, a research study was conducted in 2021 to analyze family factors associated with stunting risks in Yogyakarta Province. The study aimed to assist the government in addressing the issue and achieving the target. In this research, a hierarchical clustering algorithm using the Ward technique and a non-hierarchical clustering algorithm using the Fuzzy C-Means (FCM) approach were applied. The optimal number of clusters was determined using the average distance and figure of merit approach. Stability validation, which also used the average distance and figure of merit approach, demonstrated that the best results were achieved by the non-hierarchical clustering algorithm employing FCM. As a result, six clusters were identified: cluster 1 with 5 sub-districts, cluster 2 with 18 sub-districts, cluster 3 with 21 sub-districts, cluster 4 with 17 sub-districts, cluster 5 with 14 sub-districts, and cluster 6 with 3 sub-districts.

Keywords

• Cluster
• Stunting
• Hierarchical Clustering
• Non-Hierarchical Clustering
• Ward
• Fuzzy C-Means