Nonlinear effects of residential and workplace built environment on car dependence.

The relationships between the built environment (BE) and car dependence have been thoroughly evaluated, with a primary focus on the residential BE; however, the effects of the BE at workplaces have remained largely unexplored. Little is known about the potential nonlinear effects of the BE at both locations. Using data from a household travel survey in Changchun, China, we aimed to reveal the nonlinear effects of the residential and workplace BE on car dependence by building a gradient boosting decision trees model. The results show that the BE at both locations has strong explanatory power for car ownership and car purchasing intention. With relative contributions values of 17.90% and 18.13%, respectively, the BE at workplaces contributes less to explaining the two dependent variables than the BE at residences. All BE attributes show nonlinear effects on car ownership, and car purchasing intention and the effects differ between residential and workplace locations. • This study examines the effects of the built environment (BE) on car ownership and car purchasing intention. • The BEs at both residential and workplace locations show impacts on two dependent variables. • The relative contributions of the BE at residences are larger than the BE at workplaces. • The effects of all BE attributes at both locations are nonlinear. [ABSTRACT FROM AUTHOR]