The influence of the neighbourhood environment on peer-to-peer accommodations: A random forest regression analysis

Abstract

While the past research concerning peer-to-peer accommodation locations highlights the importance of tourist attractions, accessibility and infrastructure, investigations of whether these neighbourhood environments equally affect different Airbnb room types are limited. To fill this gap, this paper uses spatial autocorrelation and random forest regression methods to extend the research concerning Airbnb locations. The results show that the spatial distributions of all three types of peer-to-peer accommodations exhibit a distinct core-periphery pattern. Airbnb listings reflect different needs in different neighbourhood environments by providing entire homes/apartments, private rooms, or shared rooms, with entire homes/apartments involving greater environmental needs and being influenced by a combination of population density, transport accessibility, tourist attractions, and the business environment. Furthermore, we find that although the influence of the environment on the three types of Airbnb listings differs, the form of the effect is broadly the same. All three types of Airbnb listings increase and decrease in response to changes in their surroundings. This research is important in the context of understanding the expansion of peer-to-peer accommodations in cities and urban housing management.