MACHINE LEARNING AND SPATIAL DATA: PREDICTING LOCAL HEALTH VULNERABILITY

Background

In Africa, climate change is felt more severely within urban informal settlement communities. Traditional demographic techniques usually have difficulties in reflecting the changing environmental threats.

This mini course shows how Urban AI and satellite imagery (Sentinel-2) can be used to forecast potential health risks in advance. By applying machine learning models such as Random Forest and XGBoost, participants learn how to convert raw spatial data into actionable insights.

Goals

• Integrate demographic analysis with environmental modeling

• Use Sentinel-2 open satellite data for urban health monitoring

• Demystify Urban AI for public health and climate resilience

• Transform spatial data into actionable policy evidence

Learning Outcomes

• Master spatial data fusion using R and Python

• Build and validate ML models (Random Forest, XGBoost)

• Generate high-resolution vulnerability and hotspot maps

• Communicate spatial insights using professional map-based visuals

Requirements

• Basic familiarity with R or Python

• Understanding of geospatial concepts (coordinates, layers)

• Laptop with at least 8GB RAM

• RStudio or VS Code installed

• Experience with demographic or health datasets is an advantage

• Interest in urban resilience and predictive analytics