Towards a Connective, Sustainable Transportation: A Study of Relationship between Bike Sharing Service and Public Transit in Minneapolis-St. Paul Area

Abstract

Public transit offers many socioeconomic and environmental benefits but often suffers from the first/last-mile problem. Bike sharing service is designed and expected to provide first/last-mile access to transit, have a positive integration with public transit, and together contribute to a connective and sustainable transportation ecosystem. However, the relationship between existing bike sharing service and public transit is complex and ambiguous. This thesis proposes a data-driven framework with procedures and methods to investigate the competitive and complementary relationship between bike sharing and public transit systems. It defines relationships analytically and uses the criteria to detect pairs of bike sharing and transit trips correspondingly. Then, it examines the properties of paired trips and possible reasons. The thesis applies this framework to the Nice Ride bike sharing service and Metro Transit system in the Minneapolis-St. Paul Area, as a case study. The results suggest that competitive relationship exists, but only constitutes a small portion of all bike sharing trips when the spatio-temporal criteria are strict. The study of complementary relationship detects the potential first/last-mile bike trips and suggests that complementary relationship may exist and have unique spatio-temporal patterns. The correlation between bike sharing and transit ridership does not show significant competitive or complementary relationship in general, suggesting that these two systems tend to operate relatively independently from each other in the Twin Cities. However, evidence for competitive relationship can be found in several small regions. The results provide novel insights into the complex interactions between bike sharing and public transit systems and can support operation and planning practices. Since the relationships are purely defined using ridership data, we need to integrate more data to further validate our method in the future.