Non-linear relationships between built environment and commuting duration of migrants and locals

Internal migrants residing and working in cities in developing countries may suffer multiple disadvantages. Disentangling the complex relationships between built environment and their commuting duration is essential for helping them integrate into urban life, thus facilitating building sustainable and inclusive cities. However, few studies have specifically examined how built environment influences commuting duration of internal mi-grants. Moreover, prior studies frequently assume a pre-specified (linear or generalized linear) relationship between commuting duration and built environment. Employing Xiamen, China, as the case and using the random forest method, this study comparatively explores the refined non-linear effects of built environment on commuting durations of internal migrants and locals, controlling for the socio-demographics. It reveals that (1) all the built environment variables have salient non-linear and threshold effects on commuting durations of both groups; (2) built-environment effects on commuting duration differ notably between the two groups in shape, gradient, and/or threshold values; effects of built environment variables near residence and workplace also present different patterns; (3) built environment has smaller effects on migrants' commuting durations than on locals. These findings can enlighten policy makers with more nuanced and targeted intervention strategies for lightening migrants' and locals' commuting burdens.