thesamokoosi/predicting-household-poverty-nigeria

Predicting household poverty in Nigeria using R and Machine Learning (NDHS Dataset). # Predicting Household Poverty in Nigeria (NDHS 2024) This project applies Machine Learning algorithms to the Nigeria Demographic and Health Survey (NDHS) data to classify households into Poor, Middle, and Rich tiers. ### 📊 Key Findings * **Best Model:** Random Forest (3-Class Classification) * **Accuracy:** 77.4% * **Kappa Score:** 0.626 (Substantial Agreement) * **Top Predictor:** Electricity Access (Infrastructural Poverty) ### 📂 Repository Structure * `/scripts`: Final R code for cleaning and modeling. * `/visuals`: Plots showing the North-South poverty divide and model performance. * `/documentation`: Detailed Chapter 4 write-up in LaTeX format. ### ⚠️ Data Privacy Note The NDHS dataset is restricted. Only authorized users can access it from the DHS website. Therefore, the raw `.sav` file is not included here to respect privacy terms.