This study investigates the influence of hyperlocality features – specific location-related attributes – on apartment valuation. Utilizing a dataset comprising 102,885 apartment transactions in Oslo from 2017 to 2023, we employ two distinct approaches: linear models (hedonic regression and structural equation modeling) and nonlinear models (eXtreme Gradient Boosting). Our results show that proximity to amenities such as libraries, museums, theaters, and gyms, as well as higher Walk Score ratings, is associated with higher property prices. Nonlinear models outperform linear models across all performance metrics. We find that the contribution of hyperlocality variables depends critically on the level of spatial aggregation. These variables improve model performance when spatial controls are coarse but provide limited additional value when fine-grained spatial controls are included. This pattern suggests that hyperlocality measures partly act as proxies for underlying spatial structure rather than purely independent amenity effects. These findings clarify when distance-based amenity measures meaningfully improve automated valuation models, highlighting their role as substitutes for fine-grained spatial information in data-constrained environments. The results have implications for AVM design, urban analysis, and the use of spatial data in real estate valuation.
Hammervold et al. (Wed,) studied this question.