Research On Resident Travel Survey Based On Error Control Using Mobile Phone Data

With the rapid progress of urbanization,most cities in China have been transitioning from a phase of rapid growth to a stage of high-quality development.In the new development stage,the new requirement of improving the quality of urban space to meet the activity demand of residents under the condition of limited urban space resources is increasingly emphasized.Moreover,under the people-oriented governance mode of the new development stage,the rapid development of information and communication technology has promoted the improvement of resident travel survey methods.However,several key issues should be resolved in this filed,such as how to take the advantages of big data(e.g.mobile phone data)in residents travel surveys,and how to integrate traditional resident travel survey data and big data,in order to improve the accuracy of travel demand survey analysis.This study mainly aims to propose a resident travel demand survey and analysis framework based on mobile phone data and sample data from the respective of error control.First of all,based on the theoretical basis of transportation science,human geography,behavioral science,spatial statics,this study uses clustering analysis,spatial analysis and time series analysis to propose a travel demand analysis framework based on mobile phone data error control.In the travel trajectory information extraction,the grid-based re-sampling method,activity point extraction based on Mean Shift technique,and the residence and workplace identification method are proposed.In the individual behavioral indicator construction,two types of indicators are proposed to describe the spatiotemporal behavior of individual,including travel distance,distance,duration and frequency of trips,distance between activity points and residence,and directional special distribution of activities.In the spatiotemporal characteristics analysis,the spatial characteristics are analyze based on kernel density estimation and Moran’s I,and the temporal characteristics are analyze based on dynamic time warping and K-means clustering.The results show that mobile phone data can provide a high spatiotemporal resolution perspective for travel demand analysis.The proposed travel trajectory information extraction method can effectively avoid the impact of positioning errors on the individual travel information extraction.Then,the trip-based and activity-based indicators can realize the multi-dimensional continuous observation of individual travel behavior.Also,the spatiotemporal rules of travel demand reflected by the mobile phone data are basically consistent with empirical facts.Secondly,this study takes the individual spatiotemporal behavioral indicators and the structural characteristics of spatial interaction network based on travel demand as the link between mobile phone data and travel survey data,and proposes a comparative analysis framework of travel demand based on error control.Specifically,this study explores the appropriate scale,indicator,and method of comparative analysis of various behavior pattern based on two data sources,and the correlation analysis and Gaussian Mixture Model are applied to measure the similarity.Moreover,two spatial interaction networks are built based on mobile phone data and travel survey data,and the structural characteristics of network based on network analysis are used to reveal the travel demand pattern and the urban spatial structure.Node centrality indicators are used to described the local spatial activity strength and community detection method is used to identify the urban spatial form.The results prove that the results of travel behavioral indicators based on the two data sources can be comparable and the correlation analysis method and Gaussian mixture clustering method can be used to analyze the consistency of the results and the potential errors.The comparative analysis method based on the travel network structural characteristics can reflect the urban spatial structure,which provide evidences for travel demand analysis.Finally,an empirical study is made for Nanchang city using the proposed travel demand survey and analysis framework based on integration of mobile phone data and sample data,in order to demonstrate the reality,validity and effect of error control of the proposed methodological framework.Based on the proposed methodological framework,this study mainly analyzes the individual spatiotemporal behavior characteristics and the spatiotemporal pattern of travel demand.Further,this study discusses the differences between different data source and their mechanism from the perspectives of individual travel pattern and spatial interaction network,and explores the limitations of mobile phone data and sample data,and the precision promotion of travel demand analysis through data fusion.The results indicate that the travel demand analysis based on mobile phone data is reliable.For example,the degree of spatial agglomeration of workplace is higher than residence,the commuting is an important part of daily activities,the daily activity space is near the residence and is no-directional.Additionally,the statistical tests show that the indicators related to distance based on the two different data sources are highly consistent,and the results vary greatly outside the central city,especially in the suburbs.Plus,the results based on mobile phone data are always overestimated.In the comparative analysis of travel network structural characteristics,the sample survey data can hardly reflect the travel demand between some areas,thereby affecting the spatial interaction network structure.Therefore,mobile phone data has a more obvious advantage in the macro analysis of the urban spatial structure.Taking residents’travel demand as the research object,this study proposes a novel a resident travel demand survey and analysis framework from the perspective of error control.The empirical results show that the mobile phone data can reveal the spatiotemporal characteristics of travel pattern on a macro-scale to provide evidence for the travel demand and urban spatial structure analysis,and the sample data can relatively precisely characterize the individual travel behavior.Based on the proposed methodological framework and empirical results,the ultimate goal of this study is that use mobile phone data and travel survey data to improve the accuracy of travel demand estimation and analysis,in order to provide theoretical and methodological reference for the quantitative analysis of transportation science and related field.In the practice,the methods proposed in this research were applied in the Travel Survey Guangzhou 2018,Travel Survey Xiamen 2019,and Travel Survey Foshan 2020 and have been tested and optimized through theses travel surveys in different cities.Moreover,these travel surveys were successfully completed,which indicate that the proposed travel survey and analysis method based on mobile phone data is reliable and general.