Demand Forecast Of Shared Bicycle Connecting Rail Transit Stations Based On Source-sink Grid

Shared bicycles are the product of the common development of the Internet and the sharing economy in the new era.It has developed rapidly in major cities by virtue of its flexible,convenient,efficient and convenient,green and environmentally friendly advantages,and a clear positioning to solve the "last mile" problem.It has become an indispensable part of urban transportation.However,with its rapid development,many serious problems have also followed,among which the problem of the backlog of shared bicycles and the imbalance of time and space supply and demand are the most prominent.This phenomenon is especially obvious around rail transit stations.A large number of shared bicycles are backlogged near the stations,while other areas are often without cars.This not only takes up urban space,disrupts public order,affects people’s living environment,but also causes the waste of resources and reduces the efficiency of the system.In Beijing,this situation often occurs around a large number of rail transit stations.In order to analyze and solve this problem,this paper constructs a grid of shared bicycles connected to the source and sink based on a complex network,and uses this as a unit to study the demand forecast model of shared bicycles connected to rail transit.This article first takes Beijing as the research area to obtain rich multi-source data such as shared bicycle travel data,rail transit network data,road traffic network data,urban POI distribution data,and climate data,and conduct data fusion processing and statistics.analysis.Then by defining the connection area and service area of rail transit for shared bicycles,the O and D data of shared bicycles used for rail transit connection are extracted,and combined with Tyson polygons,the source and sink grid of shared bicycles connecting rail transit is constructed.It provides a suitable research unit for studying the refined management of shared bicycles.In order to combine the relevant attributes of rail transit with the demand forecast based on the source and sink grid,this paper classifies rail transit stations using a clustering method,and then substitutes the rail transit classification attributes into the demand forecast model for shared bicycles.Finally,by analyzing the relevant influencing factors of shared bicycle connection demand,combined with two machine learning algorithms of random forest and geographic weighted regression,a forecast model of shared bicycle connection demand is established based on the source and sink grid.In order to obtain a more accurate demand forecasting model,this paper designs a random forest and geographic weighted regression combined model to separate the time and space characteristics,so as to optimize the demand forecasting model for shared bicycle connections,and analyze its application through examples.