Comparison of Elbow, Silhouette and DBI Methods for Clustering Nutritional Status of Toddlers Using K-Means Clustering

Nutritional status is a measure of a person's health condition as seen from the food consumed. The implementation of nutritional monitoring or nutritional surveillance at the Community Health Center begins with data collection and reporting of toddler nutritional status with the Electronic Community-Based Nutrition Recording and Reporting System (E-PPGBM), the recording process must input toddler data from each integrated health post so that the process takes a long time until each integrated health post has measured each toddler. Furthermore, the data is processed and analyzed with Excel to produce a health information report in the form of toddler nutritional status data from each integrated health post which is reported to the health office. In the final process, namely dissemination, namely decision making for follow-up actions to carry out nutritional improvement programs. The nutritional surveillance process still has motivational constraints and performance becomes less than optimal, such as late reporting and coaching because all implementation processes are only carried out by one health center nutrition officer. The health center needs a data clustering system for the nutritional status of toddlers at each integrated health post to facilitate the decision-making process and monitoring of the nutritional health of each integrated health post so that information can be obtained to help nutrition officers prioritize integrated health posts for nutritional improvement service programs in order to achieve improvements in the nutritional health of toddlers at the Abadijaya Health Center. Evaluation of the number of optimal clusters using the Elbow, Silhouette, and Davies Bouldin Index (DBI) methods. The Elbow method obtained 4 optimal clusters, while the Silhouette method obtained 2 optimal clusters and the Davies Bouldin Index (DBI) obtained 2 clusters.