Abstract
The estimation and forecasting of travel times has become an increasingly important topic as Advanced Traveler Information Systems (ATIS) have moved from conceptualization to deployment. This paper focuses on an important, but often neglected, component of ATIS - the estimation of link travel time correlation. Natural cubic splines are used to model the mean link travel time. Subsequently, a Bayesian-based methodology is developed for estimating the posterior distribution of the correlation of travel times between links along a corridor. The approach is illustrated on a corridor in Houston, Texas, that is instrumented with an Automatic Vehicle Identification system.
| Original language | English (US) |
|---|---|
| Pages (from-to) | 53-70 |
| Number of pages | 18 |
| Journal | Journal of Transportation and Statistics |
| Volume | 7 |
| Issue number | 2-3 |
| State | Published - 2005 |
Keywords
- Automatic vehicle identification
- Gibbs Sampler
- Intelligent transportation systems
- Markov Chain Monte Carlo
ASJC Scopus subject areas
- Statistics and Probability
- Transportation