STATISTICAL ANALYSIS AND VALIDATION OF MULTIPOPULATION TRAFFIC SIMULATION EXPERIMENTS

Computer simulation has become a very powerful decision aid for varied facets of traffic engineering. Simulation experiments are often used to fit a metamodel of interest between the mean response and a selected set of input factors. This is done by carefully designing statistical experiments under alternative system designs, which are referred to as multipopulation simulation experiments. Validation and statistical analysis procedures are presented on linear metamodels from multipopulation traffic simulation networks under the common random number (CRN) strategy on three sample networks using the TRAF-NETSIM model. Under the CRN strategy, positive correlations are induced among the observations, and hence the usual statistical analysis cannot be applied to obtain point estimates and confidence intervals; therefore it must be modified. Before the statistical analysis is conducted, certain assumptions of the CRN strategy should be validated--those that, if violated, render the modified statistical analysis invalid.

Language

  • English

Media Info

  • Features: Figures; References; Tables;
  • Pagination: p. 35-44
  • Monograph Title: Traffic flow theory and characteristics with applications for intelligent transportation system technologies
  • Serial:

Subject/Index Terms

Filing Info

  • Accession Number: 00715623
  • Record Type: Publication
  • ISBN: 0309062039
  • Files: TRIS, TRB
  • Created Date: Jan 19 1996 12:00AM