Mapping and Prediction of Stunting Potential Using K-Means Clustering and Support Vector Regression (Case Study of Palopo City)

This research uses the K-Means algorithm to group cases of stunting and the Support Vector Regression (SVR) algorithm to forecast stunting patterns in Palopo City. Data from 9 districts was analyzed using K-Means clustering to find areas with high and low risk of stunting. The analysis showed a Silhouette score of 0.63 for cluster = 2, suggesting distinct groups. The optimized SVR model, using GridSearchCV, predicted 159 TB/U Short cases and 78.6 TB/U Very Short cases for 2024, with a combined total of 237.65 stunting cases. Evaluation metrics indicated that there were low error rates, with Mean Absolute Percentage Errors (MAPE) of 6% and 19% for TB/U Short and TB/U Very Short, respectively. Furthermore, the system evaluates a child's nutritional health by determining their height-for-age Z-score. If stunting is identified, the system offers practical suggestions for prompt intervention and care. This study provides a scalable model for reducing stunting, allowing policymakers to distribute resources efficiently and execute focused health initiatives.