Support Vector Machine-Based Classification of Toddler Stunting in Bandarharjo

Stunting is a significant public health issue in Indonesia, with a prevalence of $21.6 \%$ in 2022, exceeding the WHO threshold of under $20 \%$. This condition, resulting from chronic malnutrition, impairs physical growth and necessitates early detection for effective intervention. Machine learning is essential for this task, as it can analyze complex, high-dimensional data to uncover subtle patterns and risk factors that traditional methods might overlook. This study focuses on developing and evaluating a Stacked Support Vector Machine (SVM) model for predicting stunting risk in toddlers, and compares it with standard SVM models using linear, Radial Basis Function (RBF), and polynomial kernels on a dataset from Bandarharjo Health Center. The Stacked SVM model, which combines multiple weak learners to improve performance, showed superior results. By sequentially applying SVM models and correcting the errors of previous ones, Stacked SVM enhances overall predictive accuracy. This approach achieved $\mathbf{9 9 . 1 2 \%}$ accuracy and an F1-score of $\mathbf{9 4 . 8 7 \%}$, outperforming linear SVM ($\mathbf{9 8 . 8 6 \%}$ accuracy, $\mathbf{9 4 . 2 4 \%}$ F1-score), RBF SVM (98.68% accuracy, $\mathbf{9 2 . 5 1 \%}$ F1-score), and polynomial SVM $\mathbf{~} \mathbf{9 8 . 5 9 \%}$ accuracy, $\mathbf{9 2 . 2 2} \%$ F1-score). These findings highlight the effectiveness of Stacked SVM in improving early stunting detection and intervention, offering valuable insights for reducing stunting rates in Indonesia and informing future research.