Geographically weighted regression: Method for analysing spatial nonstationarity of geographical phenomenon

Abstract

The article deals with one of the new quantitative method used in geography, geographically weighted regression (GWR). This method is based on the premise that relationships between variables might not be constant across the study area and explores this phenomenon called spatial non-stationarity. Using the GWR technique to study voting behaviour in Czechia in the parliamentary election in 2002, it is evident that there is a significant difference between the linear regression and GWR models. The examples highlight the relevance and usefulness of GWR and show how it can improve geographical research and potentially also our understanding of geographical processes.