Journal of Property Tax Assessment & Administration


Luc Hermans


The implementation of geographically weighted regression (GWR) in automated valuation models (AVMs) has proven to enhance both precision and accuracy in many different study areas. However, The Netherlands has been a research area in only a couple of studies of the implementation of GWR in AVMs. This paper aims to fill this void by continuing the work of Hermans (2019) in hypothesizing that the application of GWR in municipalities with a high real estate density does not substantially improve model performance indicators. After a comparison of an AVM based on multiregression analysis (MRA) and an AVM based on GWR, this paper further strengthens this hypothesis by concluding that GWR does not significantly improve AVM performance in a densely populated city in The Netherlands.


Multiple regression analysis - The Netherlands; Automated valuation models (AVM)