Investigating Fire Sptial Process

Corr plot for selected vars

## `geom_smooth()` using formula 'y ~ x'

Gaps in Fire Districts

Relevant for Discussion about spatial cross validation

## Warning: Found less unique colors (5) than unique zcol values (38)! 
## Recycling color vector.

Error for Models

Regression Mean_MAE SD_MAE
Random k-fold CV: Just Risk Factors 0.4800000 1.2100000
Random k-fold CV: Spatial Process 0.5400000 2.7600000
Spatial LOGO-CV: Just Risk Factors 1.4400000 4.1300000
Spatial LOGO-CV: Spatial Process 2.1400000 8.3200000
TidyModels, Random k-foldCV: Just Risk Factors 0.6814349 0.2846252

Does the Spatial Factor (relating fires to fires) Produce Less Clustering Errors/

if morans I is high for errors, then we know there is some spatial process not accounted for in our model so by coomparing the morans i of no spatial process model with the mornas i of the error from the model that does account for spatial, we can see if adding the spaital feat actually made a difference

Clustering of Erros Among Three Models

Regression Morans_I p_value
Spatial LOGO-CV: Just Risk Factors -0.0156873 0.240
Spatial LOGO-CV: Spatial Process -0.0322954 0.390
Tidy, Random K-Fold: Just Risk Factors 0.0119569 0.001