Table 8. Multivariable analysis results for INLA using a zero-inflated convolution models with negative binomial errors for predicting dog rabies cases in South Africa between 2008 and 2012 using datasets with and without Kruger National Park.
| Variables | Dataset with KNP | Dataset without KNP |
|---|---|---|
| Dog population | - | -0.196 |
| PC2 | - | -2.189 |
| Temperature | 0.417 | 1.171 |
| Precipitation | 0.124 | 0.277 |
| Spatially structured residual | 1858.3 | 1866.7 |
| Non-structured residual | 1904.7 | 1764.2 |
| WAIC | 257.1 | 193.6 |
INLA: Integrated nested Laplace approximation; KNP: Kruger National Park; PC: Principal component; WAIC: Widely applicable information criterion