Modelling Spatial Heterogeneity: a Local Approach or a Global Approach?

GWR, in nature, is a local modelling approach in that it models spatial heterogeneity by sequentially fitting a series of weighted and localised regression models, centred upon each fit location. The weights are inversely proportional to the geographic distance between the fit points and the observations. SVC, on the other hand, is a global modelling approach in that it regards the spatial heterogeneity in regression coefficients as a multivariate Gaussian process and calibrates only one model rather than n localised models as in GWR.