Research On Data Mining And Analysis In Transportation Based On Mobile Communication Location

With the rapid development of urbanization, the traditional method of datacollection by artificial investigation had been unable to meet the needs of moderntraiffc operation and management. With the gradually construction and improvementof traffic information technology facilities，more and more intelligence informationgathering tools could provide support for traffic management decisions. Now mobilephone had become a widely used communication tools. It used the TDR/CDR data toachieve Cell-ID location method and record users’ continuous travel track. Comparedto other data sources, mobile location data had many advantages like low acquisitioncosts，large sample size, wide coverage, real-time and so on. As a new way to acquiredata under the IT, there is no doubt that the analysis and application of mobilelocation data could provide rich data support rang from the macro statistics to themicro behavior. Despite it had a bright future, there were still some problems in thecurrent stage:1) Currently most of the theoretical research had been done in idealenvironment of simulation or Cell-ID proactive location. It would be difficult toacquire mobile location data in the actual network environment;2) The accuracy oftraiffc travel parameters extraction still needed to be improved. The location data hadcharacteristics of no fixed cycle and large intervals limited to the Cell-ID locationaccuracy and the rule about time update;3) Large intervals and discrete characteristicslead to a lack of effective path matching method, this serious hindered the applicationof traiffc operating status monitoring;4) Lack of research on the spatial and temporaldistirbution of traiffc demand, at present, all applications for mobile location data stillconcentrated in traditional traffic travel and the state of traffic operating. The mobilelocation data were to be further excavated in spatial and temporal characteristics.Against the background，this paper has analyzed the characteristics of mobilelocation data, and proposed relevant traiffc information extraction and analysismethod.Using the improved fuzzy pattern recognition model，the method of identiyfingthe user!s work and living places based on mobile location data has been proposed.Against the large time interval for mobile location data, the unfixed update cycle andping-pong switching，the method also used the weighted mean center of multi-sectorsinstead of a single sector to characteirze the method to identify users，work and living places. It effective solved ping-pong switching problems. Through comparativeanalysis of the accuracy based on the national census data and economic census data，the method had been proved to be high practicality.By studying the communication events in subway, this paper proposed a methodto calculate the path of passengers using normal mobile position update rule.Meanwhile, against the missing date of communication event, it also proposed a pathcorrection algorithm based on the normal communication event such as phone calling，phone called，send and receive messages. The feasibility of the algorithm had beentested and analyzed by the way of artificial intentions survey. The oirginal anddestination places for date collection were Liujiayao station and Lingjinghutongstation. The result showed this method had strong feasibility.According to the difficult of matching path for mobile location data and thecharacteristics of data collected when the users in the system were moving on the road,this paper proposed LAC sequence matching method to solving path matchingproblem. On this basis，it established a model of calculating road traiffc operatingspeed. The feasibility had been tested by floating cars’ GPS data and the factors whichwould affect the accuracy had been analyzed and demarcated. The result showed ithigh accuracy.In terms of analysis on spatial and temporal distribution of traiffc demand, thispaper had analyzed the distirbution characteirstics of rail users’ work and living places，spatial and temporal distribution of population movements and checked bus&railroute choice models using the trajectory tracking features of mobile location data. Itprovided new ideas of mobile location information in traiffc applications.This paper systematic analyzed the characteristics of mobile location data andsummarized the research and application in this ifeld at home and abroad. Then itproposed the method to identify users’ work and living places，calculate the path ofrail passengers, calculate road traffic operating speed and analyze the spatial andtemporal distribution of traffic demand. The research provided a new way to collectthe data of modern traiffc information, more data support for transportation planningand management person and a new research orientation for transportation planningtheoretical model.