Traffic variables are known to be correlated over time and space as a result of traffic flow propagation. However, the correlation pattern is still largely unknown, and most of the research in short-term travel time prediction, demand forecasting, and network modeling either ignores or assumes correlation. In this paper the patterns of spatial and temporal correlations of average point speeds in a freeway setting were investigated. Five-minute speed aggregates were obtained for two directions of an urban freeway along a 12-mi segment in Los Angeles, California. Other variables included traffic flow, ramp locations, number of lanes, and level of congestion at detector stations. Weighted least squares multivariate linear regression models were fitted to the data from three different times of day (morning, midday, and afternoon) along a shorter, 6.5-mi stretch of the I-10 eastbound freeway. Estimated regression models indicated that an increase in spatial or temporal distance or both reduced the expected value of Fisher Z (transformed correlation). The positive parameters of spatial and temporal distance interaction terms showed that the reduction rate diminished with spatial or temporal distance. Higher congestion tended to preserve higher correlation; variations in road geometry carried relatively small corrections to the models. Models were cross validated on two locations: the remaining 5.5-mi stretch of I-10 eastbound and the 6.3-mi segment of I-10 westbound. Cross validation results showed that the models retained 75% or more of their original predictive capability when applied to independent samples. The developed regression models are thus transferable and are apt to predict correlation on other freeway locations.
Spatiotemporal Link Speed Correlations
Empirical Study
Transportation Research Record: Journal of the Transportation Research Board
Transportation Research Record: Journal of the Transportation Research Board ; 2390 , 1 ; 34-43
2013-01-01
Article (Journal)
Electronic Resource
English
Spatiotemporal Link Speed Correlations: Empirical Study
Online Contents | 2013
|Urban Railway Network Traffic Prediction with Spatiotemporal Correlations Matrix
Springer Verlag | 2016
|Urban Railway Network Traffic Prediction with Spatiotemporal Correlations Matrix
British Library Conference Proceedings | 2016
|