Maximum likelihood estimation of an origin-destination matrix from a partial registration plate survey

doi:10.1016/0191-2615(94)90003-5

Transportation Research Part B: Methodological

Volume 28, Issue 4, August 1994, Pages 289-314

https://doi.org/10.1016/0191-2615(94)90003-5 Get rights and content

Abstract

Previous techniques for analysing partial registration plate data are firstly reviewed. These generally fall into one of two broad categories: statistically based methods for single origin, single destination problems; and simple-minded, deterministic approaches using vehicle passage time data (i.e., the times at which vehicles pass the observation points), for surveys with multiple origins and destinations. The present paper addresses the problem of finding a maximum likelihood estimator (MLE) of an origin-destination matrix and of journey time statistics—when passage time data are available, in the multiple origin/destination case. These estimators possess the well-known large sample properties of asymptotic unbiasedness, normality, and efficiency. The proposed approach also has the advantage over the deterministic methods (used in most existing registration plate matching packages) of simultaneously analysing all possible matches between all origins and destinations, rather than considering them in some arbitrary, priority order. Since the MLEs cannot be obtained analytically, alternative numerical techniques for determining them are evaluated, with respect to their convergence properties and computational efficiency. The most appropriate of these (based on a general-purpose statistical algorithm for “missing data” problems) is described in greater detail, including issues relevant to its computer implementation. Selected results from a more comprehensive simulation study are used to illustrate the performance of the maximum likelihood approach. In the (limited) results reported, the MLEs are seen to have considerably smaller mean square errors than the deterministic methods mentioned above, but are only marginally superior to the estimators produced by an efficient heuristic technique proposed previously by the author. Further empirical work would, however, be required to establish that the patterns observed in these simulations are examples of more general phenomena. Finally, possible extensions to the method and future research directions are discussed.

References (28)

P.J.H. Shewey
An improved algorithm for matching partial registration numbers
Transpn. Res. B
(1983)
H.J. Van Zuylen et al.
The most likely trip matrix estimated from traffic counts
Transpn. Res. B
(1980)
D.P. Watling et al.
A statistical procedure for estimating a mean origin–destination matrix from a partial registration plate survey
Transpn. Res. B
(1992)
R.A. Boyles
On the convergence of the EM algorithm
J. Roy. Statist. Soc. B
(1983)
R. Brenner et al.
A general method for estimating through traffic in a road network
Highway Research Abstract
(1957)
Colin Buchanan and Partners
Micromatch Number-Plate Matching Suite
(1986)
D.R. Cox et al.
Theoretical Statistics
(1974)
A.P. Dempster et al.
Maximum likelihood estimation from incomplete data via the EM algorithm
J. Roy. Statist. Soc. B
(1977)
A.W.F. Edwards
Likelihood
(1972)
R. Evans et al.
A method for analysing partial registration-plate data
Traf Eng & Ctl
(1993)

R. Fletcher et al.

A rapidly convergent descent method for minimisation

Computer Journal

(1963)

S.I. Gass

Linear Programming

(1964)

E. Hauer

Correction of license plate surveys for spurious matches

Transpn. Res. A

(1978)

F.L. Hitchcock

Distribution of a product from several sources to numerous localities

J. Math. Phys.

(1941)

Cited by (44)

Route flow estimation based on the fusion of probe vehicle trajectory and automated vehicle identification data
2022, Transportation Research Part C: Emerging Technologies
Citation Excerpt :
Additional assumptions are needed to overcome the non-uniqueness in route flow (Bar-Gera, 2010; Wu et al., 2015). In the data-driven approach, AVI data from plate scanning devices have been frequently used (Watling, 1994). However, prior information such as OD matrix is required for constructing the objective function in order to evaluate the estimation accuracy.
Accurate estimation of route flow is crucial to traffic control optimization. Existing route flow estimation models are typically built upon strong assumptions on traveler route choice behavior or turn to data-driven approaches which heavily rely on prior OD estimation and high sensor coverage rate. This study proposes a route flow estimation framework by fusing the probe vehicle trajectory and automated vehicle identification (AVI) data. The notion of probe vehicle route penetration rate (RPR) is utilized for the construction of a novel data-driven objective function. Sensor-OD flow and observable route flow penetration are formulated as model constraints. The solution uniqueness is guaranteed though an iterative procedure under the principle of entropy maximization. Numerical studies are conducted based on pNEUMA, a large-scale full sample trajectory dataset. Results show that the proposed model performs better than the traditional path flow estimation model, especially when the prior flow information is not available. Sensitivity analysis on AVI sensor coverage rate, probe vehicle route coverage rate and RPR distribution demonstrates the proposed model’s applicability in practice.
Statistical inference of probabilistic origin-destination demand using day-to-day traffic data
2018, Transportation Research Part C: Emerging Technologies
Recent transportation network studies on uncertainty and reliability call for modeling the probabilistic O-D demand and probabilistic network flow. Making the best use of day-to-day traffic data collected over many years, this paper develops a novel theoretical framework for estimating the mean and variance/covariance matrix of O-D demand considering the day-to-day variation induced by travelers’ independent route choices. It also estimates the probability distributions of link/path flow and their travel cost where the variance stems from three sources, O-D demand, route choice and unknown errors. The framework estimates O-D demand mean and variance/covariance matrix iteratively, also known as iterative generalized least squares (IGLS) in statistics. Lasso regularization is employed to obtain sparse covariance matrix for better interpretation and computational efficiency. Though the probabilistic O-D estimation (ODE) works with a much larger solution space than the deterministic ODE, we show that its estimator for O-D demand mean is no worse than the best possible estimator by an error that reduces with the increase in sample size. The probabilistic ODE is examined on two small networks and two real-world large-scale networks. The solution converges quickly under the IGLS framework. In all those experiments, the results of the probabilistic ODE are compelling, satisfactory and computationally plausible. Lasso regularization on the covariance matrix estimation leans to underestimate most of variance/covariance entries. A proper Lasso penalty ensures a good trade-off between bias and variance of the estimation.
Freight Traffic Analytics from National Truck GPS Data in Thailand
2018, Transportation Research Procedia
Recently, demand of freight transport increases tremendously and cause serious impacts on traffic congestion and air quality. In 2015, The Department of Land Transport of Thailand has introduced a project named “Nationwide Confidence with GPS Onboard” to install GPS tracking system on all trucks in Thailand. This project provides new sources of data in analyzing the freight-related traffics and designing remedial measures for freight-related issues. This paper aims to demonstrate the use of GPS data in determining truck activities, estimating truck origin-destination matrix and estimating the flow of different commodities. Thailand is used as a case study to demonstrate the results.
Estimation of mean and covariance of stochastic multi-class OD demands from classified traffic counts
2015, Transportation Research Part C: Emerging Technologies
This paper proposes a new model to estimate the mean and covariance of stochastic multi-class (multiple vehicle classes) origin–destination (OD) demands from hourly classified traffic counts throughout the whole year. It is usually assumed in the conventional OD demand estimation models that the OD demand by vehicle class is deterministic. Little attention is given on the estimation of the statistical properties of stochastic OD demands as well as their covariance between different vehicle classes. Also, the interactions between different vehicle classes in OD demand are ignored such as the change of modes between private car and taxi during a particular hourly period over the year. To fill these two gaps, the mean and covariance matrix of stochastic multi-class OD demands for the same hourly period over the year are simultaneously estimated by a modified lasso (least absolute shrinkage and selection operator) method. The estimated covariance matrix of stochastic multi-class OD demands can be used to capture the statistical dependency of traffic demands between different vehicle classes. In this paper, the proposed model is formulated as a non-linear constrained optimization problem. An exterior penalty algorithm is adapted to solve the proposed model. Numerical examples are presented to illustrate the applications of the proposed model together with some insightful findings on the importance of covariance of OD demand between difference vehicle classes.
Route choice behavior in a radial structured urban network: Do people choose the orbital or the route through the city center?
2015, Journal of Transport Geography
We use a license plate survey to study route choice through the city center of a medium-sized Dutch city, in which car drivers can basically choose between the orbital and center ring. For a sample of 1397 trips, we fitted a multinomial logit regression model. According to this model, route choice is relatively little influenced by actual travel time. This corresponds with the fact that many drivers did not choose the shortest time route. Travel distance in combination with one “route type velocity” for all orbital routes, and one route type velocity for all center routes is the most decisive factor. The route type velocity indicates how fast and attractive routes of that type are (being perceived). The results support the hypothesis that orbital routes are more attractive as these routes avoid the busy city center. This effect is however partly offset by the fact that drivers also prefer routes in the direction of their destination. These direct routes are mainly center routes. The results show the importance of considering complete routes in relation to their location and not only as a set of links and nodes when modeling route choice.
Statistical approach for activity-based model calibration based on plate scanning and traffic counts data
2015, Transportation Research Part B: Methodological
Traditionally, activity-based models (ABM) are estimated from travel diary survey data. The estimated results can be biased due to low-sampling size and inaccurate travel diary data. For an accurate calibration of ABM parameters, a maximum-likelihood method that uses multiple sources of roadside observations (link counts and/or plate scanning data) is proposed. Plate scanning information (sensor path information) consists of sequences of times and partial paths that the scanned vehicles are observed over the preinstalled plate scanning locations. Statistical performances of the proposed method are evaluated on a test network using Monte Carlo technique for simulating the link flows and sensor path information. Multiday observations are simulated and derived from the true ABM parameters adopted in the choice models of activity pattern, time of the day, destination and mode. By assuming different number of plate scanning locations and identification rates, impacts of data quantity and data quality on ABM calibration are studied. The results illustrate the efficiency of the proposed model in using plate scanning information for ABM calibration and its potential for large and complex network applications.

View all citing articles on Scopus

View full text

Article preview

Transportation Research Part B: Methodological

Abstract

References (28)

An improved algorithm for matching partial registration numbers

Transpn. Res. B

The most likely trip matrix estimated from traffic counts

Transpn. Res. B

A statistical procedure for estimating a mean origin–destination matrix from a partial registration plate survey

Transpn. Res. B

On the convergence of the EM algorithm

J. Roy. Statist. Soc. B

A general method for estimating through traffic in a road network

Highway Research Abstract

Micromatch Number-Plate Matching Suite

Theoretical Statistics

Maximum likelihood estimation from incomplete data via the EM algorithm

J. Roy. Statist. Soc. B

Likelihood

A method for analysing partial registration-plate data

Traf Eng & Ctl

A rapidly convergent descent method for minimisation

Computer Journal

Linear Programming

Correction of license plate surveys for spurious matches

Transpn. Res. A

Distribution of a product from several sources to numerous localities

J. Math. Phys.

Cited by (44)

Route flow estimation based on the fusion of probe vehicle trajectory and automated vehicle identification data

Statistical inference of probabilistic origin-destination demand using day-to-day traffic data

Freight Traffic Analytics from National Truck GPS Data in Thailand

Estimation of mean and covariance of stochastic multi-class OD demands from classified traffic counts

Route choice behavior in a radial structured urban network: Do people choose the orbital or the route through the city center?

Statistical approach for activity-based model calibration based on plate scanning and traffic counts data