Travel Purpose Inference And Spatio-temporal Characteristics Analysis Based On Bike-sharing Data

Travel behavior theory is one of the important theories in transportation planning and management.As the direct cause of travel behavior,the study of travel purpose can provide a deeper understanding of individual travel behavior and provide reference for refined transportation planning and management.The study of travel purpose based on shared bicycle data can provide a deeper understanding of the travel characteristics of this travel mode and provide a reference for the management of the shared bicycle industry.Therefore,in this paper,using shared bicycle data,we first construct a model for identifying individual travel patterns based on DBSCAN clustering,and then construct a travel purpose inference model based on gravity model and Bayesian criterion by considering factors such as travel patterns,travel time,walking distance and cycling endpoint environment,and verify the accuracy of the model inference.Finally,the spatiotemporal distribution characteristics and road network distribution characteristics of each travel purpose of bike-sharing in Xi’an were analyzed.This study inferred the trip purposes of 2.85 million bike-sharing trips made by670,000 users,and the conclusions showed that(1)users with travel patterns in both morning and evening peaks accounted for 7% of all users,and these 7% of users generated17% of trips.The higher the number of days of weekly cycling,the greater the likelihood that users identified a spatio-temporal travel pattern.99% of users with 7 days of weekly cycling during the peak period identified a spatio-temporal travel pattern,and only 50%of users with 3 days of weekly cycling identified a spatio-temporal travel pattern.(2)About 28% of work trips,23% of home trips,16% of dining trips,15% of shopping trips,followed by feeder,school,medical,& recreation trips during the week;the proportion of work,home,feeder and school trips on weekends is lower than during the week,while the proportion of dining,shopping,medical and recreation trips is higher than during the week.(3)In terms of time distribution,work and home trips are single-peak,feeder trips are double-peak,and school trips are four-peak;in terms of spatial distribution,hotspots for work are mainly found in the high-tech zone and along some of the rail transit lines,while home trips show an outward expansion along the Second Ring Road,and hotspots for feeder trips are mainly located along the subway lines outside the Second Ring Road.(4)From the distribution of the road network,there are tidal riding phenomenon is more obvious road sections,and the gap between its morning and evening peak travel volume and the travel volume during peak hours is larger.The spatio-temporal distribution pattern based on the travel purpose of shared bicycle can provide data support for the construction of slow traffic infrastructure and management strategy formulation of relevant government departments,and provide reference for fine-grained placement management and service quality improvement of shared bicycle enterprises.