Bolu-Adekanmbi/Socioeconomic-Mapping

This project aims to achieve socioeconomic mapping from low-resource contexts through the use of mixed medium proxies # Socioeconomic Mapping from Multi-Modal Remote Sensing Data A machine learning-based approach to predict socioeconomic indicators (median household income, poverty rate, population) in low-resource contexts using satellite imagery, nighttime lights, and OpenStreetMap features. --- ## Table of Contents - Overview - Key Features - Project Objectives - Study Area - Data Sources - Project Structure - Methodology - Installation & Setup - Usage - Extending This Work - Results & Evaluation - Contributing - License - Citation --- ## Overview This project demonstrates a scalable framework for **socioeconomic mapping** in areas where traditional survey data may be sparse or outdated. By combining: - **Satellite imagery** (Sentinel-2 multispectral data) - **Nighttime lights** (VIIRS DNB monthly composites) - **Geospatial features** (OpenStreetMap infrastructure data) - **Population density** (WorldPop high-resolution rasters) We train machine learning models to predict key socioeconomic indicators at the Census Block Group level, validated against American Community Survey (ACS) ground truth data. **Why this matters:** In many low-resource contexts globally, traditional census data is infrequent or unavailable. This approach enables policymakers, researchers, and NGOs to estimate socioeconomic conditions using freely available remote sensing data and open geospatial repositories. --- ## Key Features ✅ **Multi-modal data fusion**: Integrates satellite imagery, nighttime lights, population data, and OSM infrastructure ✅ **Spatial feature engineering**: Calculates densities, proximities, and zonal statistics for block groups ✅ **Machine learning pipeline**: Implements Linear Regression, Random Forest, and XGBoost models ✅ **Spatial cross-validation**: Evaluates model generalization using spatially-aware splits ✅ **Reproducible workflow**: Jupyter notebooks with step-by-step data collection, preprocessing, and modeling ✅ **Google Earth Engine integration**: Automated …