Adopting a random forest approach to model household residential relocation behavior

Abstract

Although many studies have investigated the impact of residential relocation on travel behaviors, few studies examine how influencing factors affect residential relocation behavior. Using data from 903 families who have moved in Beijing (China), this study builds a random forest model to evaluate the relative importance and impact ways of the built environment, commuting distance, housing price, living area, and household socio-economic and demographic attributes on the residential relocation behavior of families with commuting workers or students. The results show that the built environment has the greatest influence on the relocation behavior, followed by housing price, commuting distance, socio-economic and demographic attributes, and the living area. In addition, this study finds that the effects of the built environment, commuting distance, and housing price on relocation behavior are nonlinear and have thresholds. This study can help urban planners and managers to quantitatively understand the impact of influencing factors on residential relocation behavior and provide reference values for planners and real estate development companies to plan and develop related land reasonably.