Finding where health outcomes
cluster — and why.
I'm Fetene Getnet, an MSc Health Data Science student at Debre Markos University. I build machine learning and spatial models on Ethiopian DHS data — from childhood stunting to institutional delivery — to help point health resources at the places that need them most.
Featured projects
ANC4+ Prediction & Determinants
Predicting completion of 4+ antenatal care visits among 6,893 women across 794 clusters, with GEE analysis of key determinants.
Childhood Stunting: Spatial Analysis (EDHS 2016)
Survey-weighted logistic regression and four ArcGIS spatial analyses of childhood stunting across Ethiopia.
Institutional Delivery Prediction (EDHS 2024-25)
ML pipeline predicting skilled birth attendance using XGBoost vs. logistic regression, with SHAP explainability and spatial residual mapping.
Recent notes
Why spatial residuals matter as much as accuracy
A high AUC can still hide geographic blind spots. Mapping out-of-fold residuals shows where a model systematically over- or under-predicts.