China
With bike sharing flourishing around the world, researches focus on the optimisation of station layouts, but seldom explore the dynamical features of residents’ bike-sharing travel habits. Land use factors probably indicate the activities and reflect the reasons behind residents’ travel habits; however, few studies provide empirical evidences, especially with big data analysis technology. Thus, we conduct a big data-driven empirical research to clarify residents’ travel patterns and the effects of land use thereon, which is based on the full system records of one typical bike-sharing city in China. K-modes clustering method is used to identify residents’ single and daily bike-sharing travel patterns. Three multinomial logit regression models are applied to exploring the selection mechanisms of the travel patterns. Eventually, we find that with the reference of ST pattern, the TPT-ENW pattern selection is most affected by the number of accommodation service places around travel destinations and one standard deviation increase will increase relative risk of this pattern by 1.57467. The MNT-SC pattern is most affected by the number of corporations and the other five single travel patterns are all significantly affected by the number of catering service places, which indicates commuting and dining are the travel purposes of these patterns. In addition, we find a multinomial logit model with individual fixed effects can explain residents’ daily travel pattern selection better than a pooled multinomial logit model. The analysis of residents’ travel routines explains the distribution of bike-sharing mobility and provides basis for bike-sharing system policy formulation.
© 2001-2024 Fundación Dialnet · Todos los derechos reservados