Decoding urban transportation: Trade-offs in mode choices using big data

Understanding the determinants of urban transportation mode choice is crucial for developing efficient, sustainable, and green transit systems in metropolitan areas. This study proposes a comprehensive framework that integrates big data and machine learning techniques, leveraging a large dataset comprising over ten million trips to investigate the factors influencing transportation mode choice under clear competition during peak hours. We examine how travel attributes, land use, network centrality, and demographics shape the choices of subway, bus, taxi, and bike-share users. Employing oversampling and interpretable techniques, our analysis reveals that travel attributes significantly influence transportation mode choices, especially for public transportation. Additionally, the marginal effects of the features on mode choice are efficiently captured. Interaction effects between travel time and cost further highlight the complex trade-offs travelers make under different choice probabilities. The findings underscore the importance of integrating diverse data sources for a holistic understanding of urban transportation dynamics.