Evaluating Robust Estimators in Geographically Weighted Regression for Stunting Analysis at the District-Level Across Java: A Focus on Outlier Handling

Setyowati, Silfiana Lis (IPB University)
Aidi, Muhammad Nur (IPB University)
Syafitri, Utami Dyah (IPB University)
Ernawati, Fitrah (National Research and Innovation Agency of Indonesia)

Zero : Jurnal Sains, Matematika, dan Terapan
Vol 9, No 1 (2025): Zero: Jurnal Sains Matematika dan Terapan

Rate remains high and lags behind neighboring countries such as Vietnam and Thailand. This slow progress underscores the need for region-specific interventions and a deeper understanding of local factors driving stunting to meet the 14% national target. This study applies RGWR, an improvement over GWR for handling outliers.  This method uses M, S, and MM estimators applied to the analysis of the prevalence of stunting among children under five the 2018 Riskesdas data across 85 districts in Java. Immunization reduces disease risk, growth monitoring detects stunting early, ARI management mitigates disease impact, parental height influences stunting risk, and working mothers improve family income and healthcare access, all contributing to reduced stunting. Given the regional variation in impact, stunting reduction policies should be spatially tailored, the MBG program should be prioritized in eastern Java regions.