Study On Methods Of Analyzing The Commuting Passenger Flow Trip Characteristics Based On Mobile Phone Data

With the continuous improvement of urbanization in China,the population is gradually concentrated in large cities and metropolitan areas,and infrastructure resources are used intensively,which enables industry and commerce to develop more efficiently.More employments and incomes are created,and regional economic level is increased.At the same time,the city's residential and productive land is constantly expanding,resulting in the continuous expansion of the city and causing a series of problems.In recent years,the population in central urban areas has gradually been evacuated,while a large number of industrial,commercial jobs are still concentrated in the central urban area,which has led to the continuous increase of commuting distance of urban residents and increased public commuting costs.The Outline of the 13 th Five-Year Plan for National Economic and Social Development(Draft)proposed that the layout and form of urbanization should be optimized and urban governance performance should be improved.The rational traffic based on commuters' travel characteristics can effectively increase public transport usage rate among residents,reduce commuting costs,and reduce urban traffic congestion,which is an effective mean to improve urban governance performance.Mobile phone data is one of the accesses to traffic information,which has become increasingly popular in recent years.With its characteristics of huge volume,low cost and authenticity,mobile data is widely used in the research of group travel characteristics.This article explores how to use mobile phone data to analyze the characteristics of urban commuter travelers,and studies a series of methods involved in the analysis process.Firstly,according to the characteristics of mobile phone data,a pre-processing algorithm is built to eliminate the noisy data that affects the trip characteristics analysis such as duplicate data,swing data,and drift data.Based on this,the method of identifying home/work locations of commuters based on mobile phone data is studied in detail.The method converts the mobile phone data into a sequence of activity points,and then determine their locations of home/work according to the visiting days and stay time of activity points.This method improves the previous method that identify the locations of home/work using the base station calls volume,which is in accord with the spatio-temporal characteristics of commuting behavior.Secondly,in order to obtain a reasonable commuting feature analysis scale,a traffic analysis zone(TAZ)division method based on mobile phone data is proposed.This method uses K-means++ algorithm twice to do cluster analysis of base station cells.Based on the information of home/work locations and traffic volume for the first time,it assigns traffic semantic to the base station cells to represent its land use properties.The second time is based on latitude,longitude,traffic semantic and peak flow,the study area(Nanjing)is divided into 120 TAZs.By analyzing the characteristics of time distribution of inflow,outflow and net inflow of typical TAZs,it is proved that this method can effectively ensure the consistency of land use characteristics and traveler characteristics in TAZs.Thirdly,two algorithms is built to extract the temporal characteristics information of the commuter.In the process of extracting commute spatial features,the commuter distance of commuters is calculated by using the great circle distance correction coefficient based on the straight-line distance of the residence and work locations.The spatial distribution characteristics of the commuter OD flow are depicted using the desired line chart.In the process of extracting commuting temporal characteristics,the existing method of using the minimum time difference of the continuous call sequence as the commuting duration is improved.The commuting time set of commuters is dynamically extracted with the dwell time threshold and the speed threshold,and the user's commute duration,start and end time is obtained by calculating the average value of the commuting time set.Through the analysis of the spatiotemporal information of commuters in Nanjing,it is found that the commuters in non-central districts is more likely to work nearby;the commuters in the central district has a high probability to commute in the medium or long distance."Two-peak" features is obvious among the commuters.Commuting time is much more concentrated in morning commuter period rather than evening commuter period.Finally,the study made a deep analysis of the commuting trip characteristics in Nanjing.Unsupervised learning algorithm was used to divide the traffic area,and the typical traffic area was selected to comprehensively analyze its distribution characteristics and travel time characteristics.It was found that the distribution location of regional land use was related to the location of the TAZs where the condition of working and residential separation was severe.The long-distance commuting behavior is more likely to occurred by the commuters living in the TAZs where is farther from the city center and close to metro stations.Neither the condition of office and residence separation nor the characteristics of commuting distance will take effect on the rush hour of commuters residing in the areas.