Estimating travel times from abundant data
Event details
| Date | 14.04.2011 |
| Hour | 12:15 |
| Speaker | Prof. Dr. Benjamin Heydecker |
| Location |
GC C3 30
|
| Category | Conferences - Seminars |
Contemporary Intelligent Transport Systems (ITS) provide abundant data that can be used to monitor, manage and control transport systems. However, the quality of the data is in general not known and the volume is too great for record-by-record investigation to be a practical proposition. In this seminar, we will see how statistical analysis can be applied to filter a dataset by treating it as a whole by fitting a discrete mixture of statistical distributions We then explore how this reflects on the individual observations. To illustrate use of this approach, it is applied to a certain dataset of travel times that is generated by matching time-stamped records of vehicle registration plates from automatic number-plate recognition (ANPR) system. Analysis of the mixed distribution leads to estimates of the error rate of the data collection methods, and by eliminating the corresponding component the travel time distribution can be estimated. This has dual importance: it shows how distributions of travel times can be extracted from ITS of this kind to provide estimates for direct use, and beyond that it shows how statistical analysis can be applied to manage the abundant data streams that we now receive from ITS.
About the speaker
Benjamin Heydecker is the Head of Centre for Transport Studies and Professor of Transport Studies at University College, London. His research lies in developing and applying mathematical and statistical analyses in transport studies. He has pursued this along three broad themes within the field: Traffic management and control, including real-time information and control systems; Transport safety, including the analysis and interpretation of road accident data; and Transport planning methods, including dynamic transport modelling. The interrelationships between these themes, and between each of them and methodologies for estimation, design and management are of special interest to him. He is a fellow of the Institution of Mathematics and its Applications and of the Royal Statistical Society. Heydecker is involved in the EU FP7 Network of Excellence for Advanced Road Cooperative Traffic Management in the Information Society (NEARCTIS), where he is investigating the research agenda for developments in ITS to promote a wide range of potential benefits.
Practical information
- General public
- Free
Contact
- Prof. Nikolas Geroliminis