Understanding human mobility and trip demand through e-bike trajectories is crucial for urban planning, environmental enhancement, and sustainable development. However, existing studies predominantly focus on shared (e−)bike trips, neglecting private e-bike trips. With the recent availability of sparse trajectory for private e-bikes, we established a novel analysis framework to reveal human mobility and trip demand in Wuhan, China. First, we propose a two-step method for extracting trip behavior from sparse trajectories of private e-bikes, involving the identification of staying areas and the generation of e-bike trips. Second, we establish a spatial random forest method to capture the nonlinear relationship between private e-bike trips and driving factors. Finally, we use the interpretable SHAP method to reveal the driving mechanisms of e-bike trips and explore the impact of various factors on these trips. The results indicate that (1) trip distances of private e-bikes follow a lognormal distribution, with an Adj. R-Square of 0.99, while trip times exhibit a Hill distribution, with an Adj. R-Square of 0.95; (2) Private e-bike trips are not commonly employed to address the first/last mile problem in public transportation and are more frequently used for daily commuting needs, with over 65% of these trips covering distances greater than 1 km or lasting longer than 5 min; (3) private e-bike trips positively correlate with the density of POIs like Hospital, School, and Transportation Station. However, compared to shared (e−)bike trips, Transportation Station Density, especially Metro Station Density, is less important for private e-bike trips; and (4) private e-bike trips are also positively correlated with Congestion Level and House Price, meaning that areas with severe traffic congestion or high housing prices tend to have more private e-bike trips. This study provides a new framework for understanding private e-bike trip patterns, also helping authorities better grasp the factors influencing e-bike trip demand.
Read full abstract