Research On The Relationship Between The Passenger Flow Of Subway Station And The Surrounding Built Environment

In recent years,Rail transit has been introduced to more and more cities to solve urban traffic problems.As for the large-scale rail transit network,due to the travel demand and convenience of the subway,the rail passenger flow is increasing,and some rail stations become more and more crowded.Therefore,the management of rail transit becomes more and more important.If we can divide all the stations into different types and then analyze these different types of stations,the management of rail transit will become more more appropriate.Urban Development mode can be divided into TOD(transit-oriented Development),SOD(service-oriented Development),and AOD(Anticipation Oriented Development)No matter what kind of development mode is,they are all related to the city functions.The different functional areas of the city are inseparable from various land use types in this area,and the station types is also related to the land use types around the stations.The first part of this article introduces the concept for clustering,and by using time series method,the time series of a day are used as indicators.And used the method of density distance to find clustering initial value,then according to the initial value,with k-medoids algorithm classificating Beijing subway,the total stations are divided into six types,residential-oriented stations,employment-oriented stations,mixed but more of residential-oriented stations,mixed but more of partial employment-oriented stations,mixed type of stations and a comprehensive hub stations.Finally,six kinds of stations are projected on the map,and combined with the whole location to observe the rationality of classification then find that the classification is very reasonable.The second part combining stations transfer amount,which was calculated by stations allocation algorithm is change to the subway,And by applying with the method of factor analysis,more than 40 variables are reduced to five factors,and with five clustering algorithm for clustering,after comparison,k-medoids showed the best results,by which the stations are divided into six categories,including five types can be figured out according to the five factors’ components,the rest type,by applying with the method of time series clustering,this kind of stations is divided into three categories.Finally,we project each kind of station to the corresponding subway line,and discuss the function of each line in Beijing subway.Finally,combined with the previous clustering results,compared the land use POI types the 500 meters around every kind of sites,and analyzed the number and proportion of POI types directly affecting the area.The distribution of POI around each type of site and the linear relationship between these distributions to site type and traffic volumn are also discussed.Then on this basis,several kinds of sites with weaker linear relationships are analyzed by nonlinear regression.Finally,linear and nonlinear relationships are established for each type of site.In this passage,we disvovered: 1.The commuting characteristics of Beijing subway are strong.2.Different clustering variables will result in different clustering results.3.After the subway stations are classified,different management methods and model-building progresses should be applied on different stations.4.POI data around the site have a linear or non-linear influence on passenger flow.