Resilience has been a widely used approach to assessing the response and recovery of transportation systems to various disruptions and stresses, such as extreme weather events. However, limited studies have investigated travel behavior resilience from the demand perspective, and even fewer have delved into its relationships with the built environment. Therefore, using a taxi trip dataset in Xiamen, China, and employing an advanced machine learning method (LightGBM), this study examines the non-linear effects of the built environment on travel behavior resilience during a rainstorm. The findings are as follows: (1) Location and development intensity are the most important variables for travel behavior resilience; (2) Salient non-linear relationships exist between built-environment variables and travel behavior resilience; and (3) Apparent interaction effects exist among some variables, especially between location and others. This study provides valuable insights for policymakers, offering guidance for targeted interventions to facilitate climate adaptation and enhancing resilience.