In this book two of the methods, geographically weighted regression (GWR) and spatial regression, are applied to data with spatial information.

GWR model works by assuming an existence of spatial heterogeneity in the data. Thus, instead of generalizing one regression equation for the entire data, it builds an equal amount of regression equation as the number of location in the data. The parameter estimation procedure used a similar method to that of weighted least square, with the weight usually being the distance between observations.

Spatial regression has a more elegant approach than GWR. Spatial regression only builds a single global regression equation that is generalized for the entire data. Spatial information is taken by assimilating it in the new parameters in the regression equation. Thus the spatial regression equation usually has more parameters than GWR in a single location.