Geographically Weighted Regression (GWR) is a local form of linear regression for modeling spatially varying relationships. GWR constructs a separate equation for each feature. What this means is that the relationships we're trying to model can and often change across the study area. For example, in our study, we might find that a high percentage of renters are an important predictor of burglary in one area of the county but a weak predictor in others.
GWR works by creating a local model of the variables or process that you are attempting to understand. It fits a regression equation to every feature in the study area. The variables of features that fall within the bandwidth of each target feature are incorporated into the equation. The shape and size of the bandwidth are dependent upon user input for criteria such as the kernel type, bandwidth method, distance...
