Description
Book SynopsisWith the proliferation of GPS devices in daily life, trajectory data that records where and when people move is now readily available on a large scale. As one of the most typical representatives, it has now become widely recognized that taxi trajectory data provides rich opportunities to enable promising smart urban services. Yet, a considerable gap still exists between the raw data available, and the extraction of actionable intelligence. This gap poses fundamental challenges on how we can achieve such intelligence. These challenges include inaccuracy issues, large data volumes to process, and sparse GPS data, to name but a few. Moreover, the movements of taxis and the leaving trajectory data are the result of a complex interplay between several parties, including drivers, passengers, travellers, urban planners, etc.
In this book, we present our latest findings on mining taxi GPS trajectory data to enable a number of smart urban services, and to bring us one step closer to the vision of smart mobility. Firstly, we focus on some fundamental issues in trajectory data mining and analytics, including data map-matching, data compression, and data protection. Secondly, driven by the real needs and the most common concerns of each party involved, we formulate each problem mathematically and propose novel data mining or machine learning methods to solve it. Extensive evaluations with real-world datasets are also provided, to demonstrate the effectiveness and efficiency of using trajectory data.
Unlike other books, which deal with people and goods transportation separately, this book also extends smart urban services to goods transportation by introducing the idea of crowdshipping, i.e., recruiting taxis to make package deliveries on the basis of real-time information. Since people and goods are two essential components of smart cities, we feel this extension is bot logical and essential. Lastly, we discuss the most important scientific problems and open issues in mining GPS trajectory data.
Table of Contents1. Trajectory data map-matching
1.1 Introduction
1.2 Definitions and problem formulation
1.3 SD-Matching algorithm
1.4 Evaluations
1.5 Conclusions and discussions
2. Trajectory data compression
2.1 Introduction
2.2 Basic concepts and system overview
2.3 HCC algorithm
2.4 System implementation
2.5 Evaluations
2.6 Conclusions
3. Trajectory data protection
3.1 Introduction
3.2 Preliminary
3.3 Trajectory protection mechanism
3.4 Performance evaluations
3.5 Conclusions
Part II: Enabling Smart Urban Services: Travellers
4. TripPlanner: Personalized trip planning leveraging heterogeneous trajectory data
4.1 Introduction
4.2 TripPlanner System
4.3 Dynamic network modelling
4.4 The two-phase approach
4.5 System evaluations
4.6 Conclusions and future work
5. ScenicPlanner: Recommending the most beautiful driving routes
5.1 Introduction
5.2 Preliminary
5.3 The two-phase approach
5.4 Experimental evaluations
5.5 Conclusion and future work
Part III: Enabling Smart Urban Services: Drivers
6. GreenPlanner: Planning fuel-efficient driving routes
6.1 Introduction
6.2 Basic concepts and problem formulation
6.3 Personal fuel consumption model building
6.4 Fuel-efficient driving route planning
6.5 Evaluations
6.6 Conclusions and future work
7. Hunting or waiting: Earning more by understanding taxi service strategies
7.1 Introduction
7.2 Empirical study
7.3 Taxi strategy formulation
7.4 Understanding taxi service strategies
7.5 Conclusions
Part IV: Enabling Smart Urban Services: Passengers
8. iBOAT: Real-time detection of anomalous taxi trajectories from GPS traces
8.1 Introduction
8.2 Preliminaries and problem definition
8.3 Isolation-based online anomalous trajectory detection
8.4 Empirical evaluations
8.5 Fraud behaviour analysis
8.6 Conclusions and future work
9. Real-Time imputing trip purpose leveraging heterogeneous trajectory data
9.1 Introduction
9.2 Basic concepts and problem statement
9.3 Imputing trip purposes
9.4 Enabling real-time response
9.5 Evaluations
9.6 Conclusions and future work
Part V: Enabling Smart Urban Services: Urban Planners
10. GPS environment friendliness estimation with trajectory data
10.1 Introduction
10.2 Basic concepts
10.3 Methodology
10.4 Experiments
10.5 Limitations and future work
10.6 Conclusions
11. B-Planner: Planning night bus routes using taxi trajectory data
11.1 Introduction
11.2 Candidate bus stop identification
11.3 Bus route selection
11.4 Experimental evaluations
11.5 Conclusions and future work
12. VizTripPurpose: Understanding city-wide passengers’ travel behaviours
12.1 Introduction
12.2 System overview
12.3 Trip2Vec model
12.4 User interfaces
12.5 Case studies
12.6 Conclusions and future work
Part VI: Enabling Smart Urban Services: Beyond People Transportation
13. CrowdDeliver: Arriving as soon as possible
13.1 Introduction
13.2 Basic concepts, assumptions and problem statement
13.3 Overview of CrowdDeliver
13.4 Two-phase approach
13.5 Evaluations
13.6 Conclusions and future work
14. CrowdExpress: Arriving by the user-specified deadline
14.1 Introduction
14.2 Preliminary, problem statement and system overview
14.3 Offline package transport network building
14.4 Online taxi scheduling and package routing
14.5 Experimental evaluations
14.6 Conclusions and future work
Part VII: Open Issues and Conclusions
15. Open Issues
16. Conclusions
