KaiPonel/css-madagascar-healthsheds

# Spatial Determinants of Health Across Madagascar's Healthsheds **CSS (Computational Social Science) Course Project — LMU Munich** This repository contains the analysis code for the paper *"Spatial Determinants of Health Across Madagascar's Healthsheds: The Role of Wealth, Climate, and Demographics"*, submitted as part of the CSS module at the Department of Statistics, LMU Munich. ## Overview The analysis examines how household wealth, climate, and demographic factors relate to disease incidence across Madagascar's 2,773 health facility catchment areas (healthsheds). It proceeds in two stages: - **Stage 1 (H1–H8):** Correlation and OLS regression analysis quantifying the relationships between the International Wealth Index (IWI), temperature variables, and health outcomes (malaria, diarrheal disease, ARI, malnutrition, maternal health indicators). - **Stage 2 (H9–H10):** K-Means clustering on environmental and infrastructural features to identify distinct healthshed contexts, followed by cluster-stratified regressions testing whether the wealth-health gradient varies across contexts. ## Repository Structure ``` ├── README.md ├── experiments/ │ ├── build_dataset.py # Pipeline: raw data → healthshed_dataset.csv │ ├── run_all.py # Run all hypothesis experiments │ ├── requirements.txt # Python dependencies │ │ │ ├── shared/ # Shared utilities │ │ ├── config.py # Paths, constants, disease category mappings │ │ ├── data_loader.py # Data loading and processing pipeline │ │ └── stats_utils.py # Regression, correlation, LaTeX export helpers │ │ │ ├── stage1/ # Stage 1 hypothesis scripts │ │ ├── h01_wealth_infectious.py # H1: IWI vs malaria, diarrhea, ARI │ │ ├── h02_h03_h04_wealth_health.py # H2–H4: malnutrition, maternal, mediation │ │ ├── h05_h06_h07_h08_climate.py # H5–H8: temperature, variability, interaction │ │ └── …