Does visual contact with green space impact housing pricesʔ An integrated approach of machine learning and hedonic modeling based on the perception of green space

Abstract

Green space, especially visual contact with greenery, is an important aspect of space quality assessment and has a significant impact on the premium of real estate. Due to the limitations in available data and technologies, existing hedonic studies mainly capture accessibility to green space (i.e., proximity), and how visual contact with green space (i.e., visibility) impacts housing prices is not well understood. This paper measures the intangible concept of visual contact with green space by taking advantage of street view images and community photos based on semantic segmentation. Then, based on hedonic model theory, a set of housing price determinants is selected. Finally, we construct two models, namely, random forest regression and geographically weighted regression models, to explore the capitalization effects of visual contact with green space on housing prices in Shenzhen, China. In our study, compared with indicators such as park accessibility and greening rate, variables of green visual contact are more important for housing prices. Moreover, the effects of visual contact with green space inside and outside the community display marked spatial variations. This work is a beneficial attempt to focus on the human scale for the actual demand of urban green planning. The findings of this paper add further knowledge to highlight the importance of visual contact with green space in street design and urban planning, which can enrich research on the amenity values of green space and the quality of public space theoretically and methodologically.