Geographically Weighted Models

When data are not described well by a global model, but data explain better by a suitably localized calibration, in this case, geographical weighted (GW) model, non-stationary spatial statistics, applies to describe local variability of data. For founding localized models at target locations, a moving window weighting technique is applied. Outputs of GW models provide a useful exploratory tool into the nature of the spatial data heterogeneity. Many of these GW models are included in the R (R Core Team 2014) package GWmodel.

• Geographically Weighted Summary Statistics

• Geographically Weighted Principal Components Analysis

• Geographically Weighted Regression

  • Geographically Weighted OLS Regression

  • Geographically Weighted Poisson Regression

  • Global and Local (geographically weighted) Random Forest