Exploring the impact of objective features and subjective perceptions of street environment on cycling preferences

Cycling is a popular outdoor activity, and researching the factors that influence cycling preferences helps improve cycling environments and enhance rider satisfaction. To address the limited integration of both objective street features and cyclists' subjective perceptions in existing studies, this study utilizes street view images to examine the impact of street characteristics on cycling preferences from a dual perspective. The research first employs the Mask2Former model to extract objective features and subsequently uses the MIT Place Pulse 2.0 dataset to compute subjective perceptions. XGBoost and SHAP models are then applied to explore the relationships between objective and subjective features and leisure cycling preferences. The key findings include the following: (1) Spatial heterogeneity in cycling preferences within central Beijing, characterized by higher preferences around the periphery and lower preferences in the center. (2) Beautiful, Vegetation, Lively, Safety, and Building significantly influence cycling preferences. (3) Both objective and subjective variables exhibit nonlinear effects with threshold effects on cycling preferences. (4) There are variations in the interaction effects on cycling preferences among different street characteristic variables. The study confirms the feasibility of utilizing street view images to evaluate factors that influence cycling preferences. This approach provides valuable insights for decision-makers seeking to improve cycling infrastructure and enhance the experience of urban streets.