Development of Stunting Prediction Features to Prevent Stunting Using Support Vector Machine (SVM) Algorithm

Stunting is a serious health issue in developing countries, including Indonesia, as it affects children's physical and cognitive development. Identifying and assessing stunting in children is challenging for many reasons, which include measurement inaccuracies, misinterpretation, and difficulties in implementing tools in the field. This research aims to develop features for stunting prediction using Support Vector Machine (SVM), with the goal of providing more accurate predictions and enabling timely interventions. These improvements are intended to decrease the prevalence of stunting as well as increase awareness and action capacity of health workers and parents. The study utilizes the “Stunting Toddler Detection” dataset, which is derived from World Health Organization's (WHO) stunting z-scores formula and contains 121,000 rows of data on age, gender, height, and nutritional status of toddlers, sourced from the Ministry of Health's Health Services. The SVM model demonstrated excellent performance, achieving an overall accuracy of 96%, making it a reliable tool for classifying nutritional status with consistent results across multiple metrics.