# Spatio-Temporal Generalized Linear Modeling
of Climatic Influences on Malaria Incidence
in Ghana
## Overview
This repository contains the complete R code
for the MSc thesis titled **"Spatio-Temporal
Generalized Linear Modeling of Climatic
Influences on Malaria Incidence in Ghana
(2020--2024)"**. The analysis applies a
Bayesian spatio-temporal Negative Binomial
model estimated via Integrated Nested Laplace
Approximation to examine the lagged effects
of nine climatic predictors on monthly malaria
incidence across Ghana's 16 administrative
regions.
---
## Repository Contents
| File | Description |
|---|---|
| `malaria_spatio_temporal_analysis.R` | Complete analysis pipeline |
| `README.md` | Repository documentation |
---
## Analysis Pipeline
The script proceeds through the following
sequential stages:
1. **Data Loading and Merging** — Malaria
incidence data from DHIMS-2 and climatic
data from Visual Crossing merged by region,
month, and year
2. **Exploratory Visualisation** — Monthly
malaria incidence and rainfall plots for
all 16 regions
3. **Lag Determination** — Pearson
correlation analysis for nine climatic
predictors at lags 0 to 3 months
4. **Multicollinearity Diagnostics** —
Two-stage procedure comprising inter-lag
correlation screening and Generalised
Variance Inflation Factor analysis
5. **Poisson and Negative Binomial Models**
— Overdispersion testing, model fitting,
and Poisson versus Negative Binomial
comparison
6. **Bayesian Spatio-Temporal Model** —
INLA model with CAR spatial prior, RW1
temporal prior, and spatio-temporal
interaction term
7. **Prior Sensitivity Analysis** — Three
prior sensitivity tests and two structural
specification tests with WAIC, DIC, and
Log CPO comparison
8. **CPO Model Validation** — Conditional
Predictive Ordinate statistics, PIT
histogram, and Kolmogorov-Smirnov
calibration test
9. **Risk Mapping** — Annual posterior
relative risk maps for 2020--2024
10. **Exceedance Probability Maps** —
Posterior probabil …