Mining Commuting Pattern And Urban Jobs-housing Balance From Multi-source Mobile Trajectory Data

Uncovering human mobility and urban spatial patterns from big data has become a hot research topic recently, owing to the widespread adoption of location sensing and urban computing technology. These big data are usually location-tagged, since they are collected either actively or passively through such location-aware devices as mobile phones, stationary sensors, and floating cars. How to best utilize these massive human activity data for various purposes has become a fundamental challenge to scientists and management practitioners in many different fields. One of the important urban issues that can be addressed with big data computing is the serious urban jobs-residential disparity resulting from the nationwide urbanization in the last decades, as the unprecedented population increase has completely changed the urban landscape as well as the spatiotemporal structure of urban dwellers’ daily activities. Severe urban job-housing mismatches in turn have resulted in a series of socioeconomic and environmental problems, such as traffic congestion, air pollution, and degradation of human perception about happiness. The main purpose of this thesis was to evaluate the capacity of big data computing to detect complex urban jobs-housing structural patterns for urban planning purposes. A novel approach to mine through massive trajectory data and incorporate heterogeneous web-based social media data into the analysis was designed to ensure and validate the quality of the detection results.Major research contents and conclusions are summarized as follows.Chapter 1 provides a comprehensive evaluation of the current domestic and abroad progress in the fields of urban jobs-housing spatial analysis, trajectory data mining, and urban computing based on big data. A survey of data-centric Smart City technologies from an informatics perspective was presented.In Chapter 2, a semantic trajectory data model to be adopted as an abstract of data characteristics was discussed. An ideal data model is the foundation of any data-centric research, reflecting the accurate understanding and precise expression of the research object. First, a review of traditional concepts of trajectory and a relevant semantic modeling method was presented. Second, a comparative analysis was conducted between the classic relational model and the aggregation model in terms of semantic modeling and expression, which led to the belief that the latter was a superior conceptual model for representing semantic trajectory structures and could easily be adapted to the complex, diverse, and large-scale characteristics of the data. Then the model building procedure was outlined. Third, MongoDB, which is an open-source, document-oriented database management system, was selected for semantic trajectory database design and implementation. Further analyses were conducted on the semantic trajectory data management, which consists of data retrieval, spatial indexing, and distributed clustering. Finally, a distributed database cluster system for aggregation-oriented semantic trajectory modeling was established by a "replica set+sharding" cluster solution and then compared with traditional relational database in data insertion, query and spatial query performance. The results showed that the aggregation-oriented semantic data model was more efficient when it worked in the NoSQL database cluster.Chapter 3 described the acquisition and processing of heterogeneous human activity data from a variety of sources. The human activity data were classified into active perception and passive perception in the thesis. The former category includes various survey data, and the latter category includes taxi trajectory and location-based social networks data. Humongous trajectory data collected from taxi companies were processed to provide large coverage and dynamic representation of human activities. A new method for extracting taxi origin-destination (OD) points was described. The acquisition and processing of spatiotemporal social media data were presented in the final section of this chapter. A Sina microblog data crawling and processing platforms was designed, and a spatial clustering method for data preprocessing was introduced as well. A procedure of coupling GPS-enabled smartphones and web-based feedback survey procedure with geospatial recalls for self-reports from respondents was designed and implemented. A complete travel survey system was described in detail, including the acquisition module, the analysis module and the feedback module. Finally, the quality of the final survey results was evaluated.In Chapter 4, several recognition methods for commuting behavior patterning based on multi-source location-aware data were presented. First, the thesis focused on trajectories obtained from GPS-enabled taxis and their applications for mining urban commuting patterns. A novel approach is proposed to discover spatiotemporal patterns of household travel from the taxi trajectory dataset with a large number of point locations. The approach involves three critical steps:spatial clustering of taxi OD based on urban traffic grids to discover potentially meaningful places, identifying threshold values from statistics of the OD clusters to extract urban jobs-housing structures, and visualization of analytic results to understand the spatial distribution and temporal trends of the revealed urban structures and implied household commuting behavior. Second, the commuting behavior patterns from the location of microblogs were identified through fuzzy pattern recognition, using the individual behavior patterns from household travel survey data as identifying characteristics. The evaluation step showed that the recognition results closely matched the observed patterns and thus well validated the proposed method.In Chapter 5, a case study of jobs-housing balance with heterogeneous human activity data from a variety of sources in Shanghai, China was presented to discover its urban spatial patterns. First, related research of jobs-housing balance and urban spatial structure as well as the related trajectory data mining methods was introduced. Second, a commuting efficiency measure, known as excess commute, was presented. Then the commute efficiency, jobs-housing balance, and excess commute in Shanghai were investigated from the microblog location data. It also compares relevant indicators of Shanghai with those in other Chinese cities. Third, the association relationship between jobs-housing balance and characteristics of urban communities was discussed. Finally, a geographic visualization method with taxi trajectory mining results was presented to explore the commuting network patterns of Shanghai.In Chapter 6, the conclusion and discussion of the dissertation was stated.