Research On The Relationship Between Bike-sharing And Built Environment Based On Feature Mining

As an important part of urban public transportation,bike-sharing provide short-distance travel and connectivity for urban residents.After experiencing the outbreak period,the growth rate gradually slowed down and entered the stage of fine management.However,due to the motor-oriented habit,the planning and management of bike-sharing is not considered in the urban planning stage.There is still obvious tidal effect,uneven spatial distribution,which seriously affect residents’ travel choices.Therefore,this paper explores the travel characteristics and feedback mechanism between urban built environment and bike-sharing from a data-driven perspective.This study mainly focused on the trip record data of bike-sharing,supplemented by the data of administrative boundary,road network,POI,population,etc.Based on the preprocessing of multi-source data,starting with the basic operating characteristics of bike-sharing by travel time,travel distance and travel intensity.It is found that bike-sharing have the characteristics of "service working days,peak trends,short-distance travel and low turnover rate".Then,the travel characteristics of hot spots during peak hours were explored,and the relationship between tidal characteristics and land use types is analyzed by using the kernel density estimation and regional hot spot detection model.Taking into account the connection function of bike-sharing,cluster analysis was conducted on the service mode of subway stations during rush hours.In addition,aiming at users’ cycling behaviors,the path planning service of Auto Navi Map is used to obtain the cycling track and analyze users’ cycling characteristics from cycling path,cycling chain and cycling purpose.In order to explore the reasons for the differences in the temporal and spatial characteristics of bike-sharing,the urban built environment element set is used as the explanatory variable to analyze the relationship between the urban built environment and cycling demand,this paper analyzes four perspectives: road network,land use,public transportation and regional attributes.The spatial correlation of cycling demand is verified,and the Spatial Lag Model(SAR)and Spatial Error Model(SEM)were compared.Moreover,building the theoretical system of bikesharing spillover effects and testing the existence of spillover effects,Space Doberman Model(SDM)is established and the direct,indirect and total effect solution is derived,the results verified the necessity of introducing spatial spillover effect.It is show that the spatial spillover effect can reflect the interaction between the built environment and the cycling demand in different regions,and the mechanism of the spillover effect of the built environment on bikesharing is established from the perspectives of endogenous force,external gravity and basic force.Finally,an integrated architecture for bike-sharing and city management systems is established.