<![CDATA[The availability of open data enables computational-based monitoring of toddler nutritional problems, but local administrative datasets are often small, aggregated, and lack target labels for supervised modeling. This study aims to (1) determine the most appropriate number of case volume clusters, (2) characterize the clusters of stunting, wasting, and underweight, and (3) build a simple and reproducible analysis baseline using Google Colab. Secondary data was obtained from the Aceh Government's open sources and consists of 15 aggregate observations from Southeast Aceh Regency between November 2022 and March 2023. The modeling variable is the number of cases in individuals. The research stages include data understanding, quality checking, descriptive analysis, K-Means modeling for k=2 to k=5, and internal evaluation using the Silhouette Coefficient, Davies-Bouldin Index (DBI), and Calinski-Harabasz Index (CHI). The results show no missing values or duplications. The highest silhouette value was obtained at k=2 with 0.8189, alongside a DBI of 0.2286 and a CHI of 150.6841. The high-volume cluster is centered at 1,182.6 individuals and contains five stunting observations, while the lower-volume cluster is centered at 545.7 individuals and contains five wasting observations and five underweight observations. The total of the three indicators changed from 2,469 individuals in November 2022 to 2,122 individuals in March 2023 (-14.05%), but the change was not monotonic as an increase occurred in February 2023. The novelty of this research lies in a minimal baseline that empirically selects k, uses three evaluation indices, incorporates temporal examination, and limits cluster interpretation to relative volumes. The results can serve as an analytic audit prototype for regional nutrition data, but cannot yet be used to infer prevalence, causality, or individual risk. The research testing can be downloaded from github.com. Keywords: Data Mining; K-Means; Clustering; Stunting; Wasting; Underweight; Google Colab]]>
Copyrights © 2026
