Examining the Nonlinear Effects of Residential and Workplace-built Environments on Active Travel in Short-Distance: A Random Forest Approach.

Environmental pollution and health problems caused by the excessive use of motor vehicles have received widespread attention from all over the world. Currently, research lacks attention to the nonlinear effects of the built environment on short-distance active travel choices. It is important to understand these non-linear correlations, because it would be more feasible and necessary to promote a shift from car users to walking and cycling mode choices over short commuting distances. A random forest model was used to analyze the nonlinear effects of residents' social characteristics and the built environment of their homes and workplaces on their choice of walking and cycling. The results show that the built environment has a greater impact on short-distance active travel than the socio-demographics attributes. Residential and workplace-built environments have equal importance and they have significant non-linear effects on both short-distance walking and cycling. The nonlinear effects of the built environment on walking and cycling differed significantly, and the study specifically revealed these effects.