Stunting Prediction in South Sumatra Province: Machine Learning Approaches Using SKI 2023
Alfensi Faruk, Dian Cahyawati Sukanda, Endang Sri Kresnawati

Department of Mathematics, Universitas Sriwijaya

Abstract

Stunting is still one of the main health issues in Indonesia, including in South Sumatra Province. To achieve zero hunger by 2030, as outlined in the United Nations Sustainable Development Goal 2, effective strategies are required to reduce the stunting prevalence. This study utilises logistic regression and gradient boosting machines, on individual-level data from the Indonesian Health Survey (SKI 2023) consisting of 2,940 children under five years of age in South Sumatra Province. The prevalence of stunting was 19.3% (weighted). Birth weight, birth length, living conditions, sanitation, and sex are the strongest factors that significantly affected the stunting status in South Sumatra Province. This result was also confirmed by the conducted SHAP analysis. Even though both models show only modest discrimination (AUROC \(\approx\) 0.59), our findings are informative to enhance the strategies of stunting prediction. The results of this work provide a potential baseline for future policy improvement.

Keywords: Stunting, South Sumatra, SKI 2023, Logistic regression, Gradient boosting machine

Topic: Mathematics and Applications