Predicting Stunting Prevalence in Indonesia Using XGBoost and Artificial Neural Networks

Stunting remains a critical public health issue in Indonesia, particularly in rural areas, where the prevalence of stunted children continues to rise. Stunting refers to a condition where toddlers have a height significantly below the average for their age, caused by prolonged inadequate nutritional intake. This condition poses severe long-term consequences, including delayed brain development, reduced cognitive abilities, and an increased risk of chronic diseases such as diabetes, hypertension, and obesity. According to the Indonesian Nutrition Status Survey (SSGI), the stunting prevalence rate in Indonesia was 21.6% in 2022, a figure that threatens the development of the nation's younger generation. To address this issue, effective monitoring and predictive models are essential for supporting government decision-making and policies aimed at reducing stunting rates. This study implemented and evaluated the performance of Artificial Neural Networks (ANN) and XGBoost algorithms as regression models for predicting stunting prevalence in Indonesia. Hyperparameter optimization was conducted using the Tree-Structured Parzen Estimator (TPE) approach to improve model accuracy. Among the tested models, the XGBoost algorithm demonstrated superior performance, achieving a Mean Squared Error (MSE) of 21.57 for training and 16.11 for testing, outperforming both ANN and Linear Regression (LR). Additionally, the computational cost analysis revealed significant differences in training times: approximately 1 minute for an ANN with one layer, 1.5 minutes for an ANN with two layers, 0.1 seconds for XGBoost, and 0.05 seconds for LR. These findings underscore XGBoost's effectiveness and efficiency, making it a promising tool for stunting prediction and policy support in Indonesia.