Citation

Gafarian AV, Halati A. Transp. Res. Rec. 1986; 1091: 29-36.

Copyright

(Copyright © 1986, Transportation Research Board, National Research Council, National Academy of Sciences USA, Publisher SAGE Publishing)

DOI

unavailable

PMID

unavailable

Abstract

Simulation models are increasingly becoming the most convenient tool for traffic studies. Users of such models need valid statistical methods to draw correct inferences. Presented in this paper is one such method applicable to several important traffic parameters. The motivation for this research arose from a study sponsored by the FHWA, U.S. Department of Transportation, to develop statistical guidelines for simulation experiments with traffic models. NETSIM, widely used for simulating vehicular traffic flow on urban streets, was used in the study. The output of the NETSIM model includes estimates of average speed, average delay per vehicle, and average travel time per vehicle mile. Because NETSIM uses the ratio of sample means to estimate these parameters, a situation exists that involves the ratios of observations that are in fact autocorrelated and cross correlated. In this paper, the efficacy of the ratio of sample means (used in NETSIM) as an estimator of the ratio of steady state means is discussed. Monte Carlo experiments have demonstrated that the user of the NETSIM model, in estimating these parameters from the model output, must apply statistical techniques based on ratio estimators. A technique that provides a measure of the accuracy of the estimate with a confidence interval is developed and demonstrated. The efficacy of the method is assessed through Monte Carlo experiments. The method is easy to use and can be applied just as readily to field data. It can be extended to the comparison of model outputs to field observations for simulation validation studies.

Language: en

Keywords

URBAN PLANNING; MATHEMATICAL STATISTICS - Monte Carlo Methods; STREET TRAFFIC CONTROL - Simulation