rappiok/Child-Malnutrition-in-Sub-Saharan-Africa

Analyzing child malnutrition and mortality trends across Sub-Saharan Africa using WHO data, with fairness auditing and causal analysis planned. # Child Malnutrition in Sub-Saharan Africa ## Overview This project analyzes child malnutrition and mortality trends across Sub-Saharan African countries (2010–2024) using real-world data from the WHO Global Health Observatory (GHO). It covers the full pipeline from data collection through cleaning, exploratory analysis, predictive modeling, and a regional disparity check. ## Motivation Child malnutrition remains a major public health challenge across Sub-Saharan Africa, and understanding its trends and drivers has real implications for policy and resource allocation. This project is also a way to build hands-on experience with real data for advancement. ## Data Source - **WHO Global Health Observatory (GHO) OData API**: who.int - Indicators include stunting, wasting, under-five/infant/neonatal mortality rates (JME: UNICEF-WHO-World Bank Joint Malnutrition Estimates). - Data is pulled live via the API rather than stored as static files, so the notebook always reflects the latest published estimates. ## Methodology 1. **Data collection** — Queried the WHO GHO API for nutrition and mortality indicators across Sub-Saharan African countries. 2. **Data cleaning** — Identified that the API returns multiple estimates for a country per year (e.g. across sex/age subgroups), and this was resolved by aggregating with the mean per country and year rather than discarding data. 3. **Missing data handling** — Dropped indicators with low coverage, interpolated remaining gaps by country (justified by the gradual year-to-year nature of health indicators), and removed the small number of countries still missing data after interpolation. 4. **Exploratory & trend analysis** — Visualized regional trends (2010–2024), ranked countries by change in stunting and mortality, and examined correlations between malnutrition and mortality indicators. 5. **Predictive modeling** — Built a Random Forest model to predict under-five mortality from stunting an …