Spatio-temporal Characteristic And Influencing Factors Of Dockless Bike Sharing Usage

Recently,the development of bike sharing in China is gradually stable.With its advantages of high efficiency,convenience,low carbon and green,bike sharing has gradually integrated into people’s daily life,especially in solving the "last mile" problem.But on the other hand,bike sharing also has many problems,such as blind delivery,random parking,inefficient scheduling and so on.Scientific understanding of the spatiotemporal distribution characteristics of bike sharing and its influencing mechanism,and full discussion of the scale effect of various factors on bike sharing are the basis of correctly grasping the rules of bike sharing,and are of great significance to the control of bike sharing and efficient spatiotemporal scheduling.Based on the multi-source spatiotemporal data such as GPS record data,POIs data,street view image data,Wechat real-time population data and road data,this study uses space time cube model,emerging hot spot analysis,Geo Detector model,full convolution network model,backward stepwise regression method and geographically weighted regression model to analyze about twenty million of bike sharing usage record data in two consecutive weeks in Shenzhen,including the research of riding behavior characteristics,spatiotemporal characteristics and spatiotemporal patterns,the interactive effects of built environmental factors on the origin and destination of bike sharing usage and the discussion of MAUP(modifiable areal unit problem),and the analysis of the relationship between holidays riding and urban greening.The results are as follows.(1)Riding behavior characteristics,space-time characteristics and space-time mode: the morning peak riding on weekdays has the characteristics of fast speed and short time consumption;the evening peak riding on weekdays and the whole day riding on weekends have the characteristics of longer distance and longer leisure time.The average riding time in Shenzhen is shorter than that in Beijing,Shanghai and Guangzhou;the riding in Guannei is concentrated in Futian,Nanshan and Bao’an;the riding in Guanwai is concentrated in Longhua.Based on the OD flow and its traffic distribution,it is found that shared bicycle riding in Shenzhen has obvious characteristics of commuting between work and residence;on the whole,the grid of cycling hot spot mode is closer to the city center and more accessible to the subway station.(2)In the aspect of interaction analysis and MAUP discussion of built environment factors of daily cycling: This study discussed the MAUP(scale effect and zoning effect)in the modeling of built environment factors’ impact mechanism on bike sharing,and finally selected the 600 m and natural break method as the most suitable grid scale and zoning method for the study of bike sharing impact mechanism in this study area,and obtained some findings of planning and bicycle scheduling: on the one hand,the built environment factors such as distance from subway station exit,POI distribution of occupational places,POI distribution of residential places,POI distribution of entertainment places,land use mix and real-time population density are sensitive to spatial scale in modeling,which means that urban planners should pay more attention to spatial scale when planning these built-up environment factors Inter scale.The results show that the relative importance of these factors is inconsistent in different grid spatial scales,which built-up environmental factors should be paid more attention to by planners in a specific spatial scale.On the other hand,through the interactive detection model,it is found that the distance from the exit of the subway station and the POI distribution factors of various facilities have a greater interactive influence on the riding of shared bicycles,but the influence of these interaction factors on the starting point and the ending point is different in different days,different periods of time.(3)In terms of the analysis of the relationship between holiday cycling and urban greening: This study obtains two urban greening indicators from the perspective of overlooking and human eyes through NDVI and street view map data,and uses GWR Model to analyze and compare the spatial heterogeneity of the impact of the two Urban Greening Indicators on daily and holiday bike sharing.The results show that the street view green vision rate of human eye perspective has a significant positive effect on bike sharing on weekdays,weekends and holidays,while NDVI has a significant negative effect on weekends and holidays,but not on weekdays.This shows that urban greening from the perspective of human vision is closely related to riding behavior,because it is more in line with people’s real feelings and experience of urban greening,which is consistent with the existing research.In order to create a riding friendly environment,we should pay attention to the street landscape greening from the perspective of human eyes in urban planning and design.