owenhaworth/Spatial-Data-Analysis

Spatial Data Analysis of Malaria Prevalence in The Gambia # Spatial Data Analysis of Malaria Prevalence in The Gambia Welcome! This repository contains my Spatial Data Analysis assignment completed at the University of Toronto. The report was created using **R Markdown** and knitted to **HTML**, allowing all analyses, maps, and tables to be viewed directly in your web browser. --- ## How to View Click the link below to open the HTML file via **GitHub Pages**. No GitHub account is required — the HTML will open like a regular website. **→ _View the full HTML report here_** --- ## About the Project This assignment explores **geospatial statistics**, **Bayesian modeling**, and **spatial prediction** using malaria data from **The Gambia**. Analyses were conducted in **R**, using packages such as **geoR**, **sf**, **INLA**, **tmap**, and **raster**. The project demonstrates: - Spatial data wrangling and visualization - Prior predictive simulation - Spatial binomial regression using INLA - SPDE approaches with Matérn and exponential covariance models - Spatial prediction and uncertainty quantification - Mapping exceedance probabilities --- # Assignment Overview ## Q1.1 – Spatial Data Preparation and Visualization ### Key Tasks - Load the malaria dataset and aggregate malaria test results by sampling location. - Convert coordinates to an **sf spatial object** with a proper CRS. - Extract elevation values (altitude) from raster data. - Produce a two-panel map displaying the **total number tested** and **prevalence** across The Gambia. --- ## Q1.2 – Prior Predictive Simulation ### Key Tasks - Compute geodesic distance matrices between sampling locations. - Simulate **100 prior predictive datasets** using Bayesian logistic regression with: - Normal priors for regression coefficients - Gamma priors for spatial range and smoothness - A Matérn covariance structure - For the first six locations, generate histograms of prior-predicted prevalence with observed values overlaid. --- ## Q1.3 – Mapping Prior Predictive Dra …