Contribution of Streetscape Features to the Hedonic Pricing Model using Geographically Weighted Regression: Evidence from Amsterdam

This paper aims to explore the contribution of streetscape features to Airbnb accommodation pricing, allowing hosts and tourists to benefit from a more transparent pricing scheme. To this end, a hedonic pricing model based on Geographically Weighted Regression (GWR) is estimated using the data in July and November 2020 in Amsterdam. With the semantic segmentation model Deeplabv3 trained by the Cityscapes dataset, the percentages of 4 types of pixels in the images sourced from Google Street View are measured. Three indicators representing streetscape design (greenery, enclosure, and walkability) are calculated and added to the models. The results suggest that greenery and enclosure have upward effects while walkability has a downward effect on accommodations’ prices. Moreover, accounting for spatial heterogeneity and the inclusion of streetscape related variables improve the model fit by increasing adjusted R-squared and reducing residual sum of squares.