Analyzing The Passenger Travel And Traffic Running Characteristics Based On Mobile Trajectory Big Data

With the rapid development of global economy and the steady progress of urbanization,the mining and analysis of mobile trajectory big data has become a highly concerned issue of people’s livelihood in academia,industry and government departments.In this paper,we establish a Hadoop distributed computing platform(big data analysis platform)based on a MapReduce parallel processing framework to realize the efficient mining and deep analysis of large-scale mobile trajectory data(the GPS data of taxi).Then,we analyze the characteristics of passenger travel,to control the trip rules of urban residents in real time.Finally,we analyze the characteristics of taxi operation for the dynamical perceive traffic conditions.It can provide the theoretical basis and decision reference for urban traffic management and control.The contents and innovative results of this paper are summarized as follows:1.Data pretreatment and big data analysis platform.Firstly,the big data analysis platform based on the MapReduce framework is built to realize the Hadoop distributed storage and parallel computation of big data of mobile trajectory.Furthermore,the big data analysis platform is verifed by using the large-scale taxi GPS trajectory data,which provides the guarantee for the testing,mining and analysis of the follow-up work;Finally,analysis the data characteristics of taxi GPS data,and the intelligent extraction and parallel preprocessing of taxi GPS data based on big data analysis platform are analyzed.2.Analysis on spatial-temporal characteristics of the passenger’s travel.Firstly,the OD point of the passenger’s travel is extracted by utilizing the big data analysis platform to conclude the peak value distribution of the passenger’s travel at each time interval.Moreover,according to the peak value distribution,the whole area of Beijing is subdivided to analyze the characteristics of the passenger’s travel in line with the difference in distribution trends on workday and day off.Lastly,clustering analysis on distribution state of the passenger’s travel is implemented by making use of the built big data analysis platform according to the map.3.Analysis of taxi operation characteristics.Firstly,massive small file processing methods(HAR,CFIF,SF)were realized by utilizing the big data analysis platform to make up for the existing defects of the traditional big data analysis platform in processing big data and file data set including “slow read-write speed,high memory consumption and low handling efficiency”.Secondly,we incorporate SF into Frequent Pattern growth(FP–growth)algorithm.Based on the MapReduce paradigm,the frequent pattern growth optimization algorithm(MR-PFP)is reconstructed by Map.Reduce and other functions to realize parallel mining of frequent itemsets with large moving trajectory data.Thirdly,MR-PFP algorithm was applied into the real taxi trajectory big data containing massive small file data set to realize spatial-temporal correlation analysis on taxi operation characteristics.Finally,according to the experimental result,it manifested the optimized MR-PFP algorithm was more efficiency and scalability than the built parallel frequent pattern growth algorithm(PFP).