Three models for the estimation of turning flows at intersections are compared. Of these, two, the Information Minimising model (IM) and the Bayesian model (B) have previously been described in the literature. The third, the Maximum Likelihood mode (ML) is new. All three models combine prior information (from historical data or a recent short manual count) on the turning flows with long-term automatic counts on the flows into and out of the junction, to produce estimates of the turning flows. Tests using a variety of data-sets show that there is a very close similarity in the estimates produced by the three models and that, in terms of a comparison of these point estimates, all three models behave equally well. Other considerations, however, show up significant differences between the models. The most important of these is that models B and ML produce standard errors of the estimates, whilst model im cannot do so. All three models are simple to program, but B is non-iterative and is easily modified to cope with different generalisations of the basic problem. Overall the Bayesian method would appear to be the most appropriate choice of model for the estimation of turning flows at intersections. (TRRL)

  • Availability:
  • Corporate Authors:

    Printerhall Limited

    29 Newmart Street
    London W1P 3PE,   England 
  • Authors:
    • MAHER, M J
  • Publication Date: 1984-1

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 00393260
  • Record Type: Publication
  • Source Agency: Transport Research Laboratory
  • Files: ITRD, TRIS
  • Created Date: May 31 1985 12:00AM