Mathematical characterization of spatiotemporal congested traffic patterns: mixed speed data analysis in the greater Toronto and Hamilton area, Canada

This paper formulates a comprehensive methodology for analyzing, quantifying and identifying congestion characteristics based on speed distribution. Utilizing vehicle speed data, a mathematical approach is applied, in order to characterize roadway segments, in terms of travel reliability, congestion...

Full description

Saved in:
Bibliographic Details
Published in:Transportation planning and technology 2016-04, Vol.39 (3), p.318-328
Main Authors: Kyriakopoulou, Natalia, Photis, Yorgos N., Kanaroglou, Pavlos
Format: Article
Language:English
Subjects:
bimodal distribution
Congestion
Data analysis
Data processing
expectation-maximization algorithm
Gaussian
Gaussian mixture model
Mathematical analysis
Mathematical models
Roadways
Segments
Speed limits
Traffic accidents & safety
Traffic congestion
Vehicles
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:This paper formulates a comprehensive methodology for analyzing, quantifying and identifying congestion characteristics based on speed distribution. Utilizing vehicle speed data, a mathematical approach is applied, in order to characterize roadway segments, in terms of travel reliability, congestion severity and duration. We argue that the Gaussian mixture model (GMM) and its parameter combination is the appropriate tool if we are to obtain quantitative congestion measures and rank roadway performance. A significant contribution of our approach is that it is based on assumptions regarding mixed components as well as speed distribution and can be applied to large databases. We test our framework on the greater Toronto and Hamilton area in Ontario, Canada, and conclude that congestion quantification through the application of the GMM can be successfully accomplished. Results indicate that speed patterns differ significantly between counties as well as days of the week.
ISSN:0308-1060
1029-0354
DOI:10.1080/03081060.2016.1142226