The next session of the ElementR group will be led by Thierry Feuillet (Université de Caen Normandie, UMR 6266 CNRS – IDEES) and will take place on
March 20, 2025
from 1pm to 4pm
room 1.122, Bâtiment Sud
Condorcet Campus
Geographically weighted regression is one of the leading spatial analysis methods for modeling the heterogeneity of statistical relationships across space. It belongs to the family of models known as “spatially variable coefficients.” Appreciated for the simplicity of its implementation and the richness of its results, it is applied to a wide range of themes and is constantly evolving owing to multiple extensions that enhance its parameterization capabilities and computational efficiency. As with any method, it has both theoretical and statistical strengths and limitations, which are useful for identifying and discussing ways to refine our geographical analyses.
This session had three objectives. First, the theoretical and statistical underpinnings of the method will be presented, focusing on its ability to incorporate spatial non-stationarity, as well as the potential for hybridization with other approaches, notably multilevel models, to better apprehend contextual effects in geography. This is followed by feedback on its empirical application to a range of themes and issues as a basis for discussion. Finally, a reproducible R script was implemented and discussed based on a concrete example to explore the subtleties of model calibration and interpretation.
Registration is mandatory: https://evento.renater.fr/survey/la-regression-geographique-ponderee-gwr-par-thierry-feuillet-3bq7shgp
