Abstract
In the urban traffic research field, taxi detour behavior analysis can be regarded as one of the most crucial and challenging topics accounting for real-world routing network dynamics with complicated external inducement such as ``avoiding congestion sections'', ``unfamiliarity with road maps'' or just ``earning more fee under a longer travel path''. We carried out an interdisciplinary research framework to build a more holistic and profound view of the spatio-temporal distribution of the taxi detour behavior at directional road segment (DRS) level. First, a map matching based detour clustering method was proposed to deal with one week of taxi GPS tracing (divided into 3.4 million occupied trips). Then we employed an established multi-layer road index system in Shenzhen, China, to illustrate the spatio-temporal distribution variation of taxi detour features and statistics. Furthermore, three categories of DRS factors related to road structural attributes, traffic dynamics and point-of-interests (POIs) were defined to fit a selected-sample-based binary logit model. Some remarkable findings include: (i) in Shenzhen on average, 23.5 percent of taxi trips made a detour larger than 2.1 kilometers, which could be astonishingly high considering that only a very few trips yielded formal complaints for fraudulent detouring; (ii) both the level of detour intensity and ratio are affected by road features and dynamics in different spatio-temporal interaction patterns.
Original language | English (US) |
---|---|
Journal | IEEE Transactions on Intelligent Transportation Systems |
DOIs | |
State | Accepted/In press - 2021 |
Keywords
- Indexes
- Junctions
- Magnetic resonance imaging
- map matching
- multi-layer road intex systems
- network complexity
- Public transportation
- Roads
- Routing
- spatio-temporal distribution features.
- taxi data mining
- Taxi detour behavior
- urban computing
- Vehicles
ASJC Scopus subject areas
- Automotive Engineering
- Mechanical Engineering
- Computer Science Applications