This paper describes procedures for aggregating disaggregate choice models after estimation of the choice model parameters to obtain an aggregated model structure. This structure consists of (a) a disaggregate choice model, (b) a representation of the distribution of explanatory variables, and (c) an aggregation procedure. A taxonomy of aggregation procedures that classifies models according to their structural characteristics is developed. Errors in prediction by use of alternative aggregation procedures are empirically estimated. Analysis of these errors leads to conclusions about the performance of different aggregation procedures. These conclusions suggest the following guidelines for aggregate travel prediction using disaggregate choice models: (a) Disaggregate choice models may be most effectively used for prediction at high levels of aggregation appropriate to policy analysis; (b) enumeration procedures should be used whereever adequate sample data are available, especially at high levels of aggregation; (c) when sample data are not available, classification procedures should be based on the most important class distinction which will be differences in choice set when such differences exist; (d) when data are not available to predict class specific variable values, predictions by the naive procedure should be adjusted for differences in choice set when such differences exist; (e) the specification of the underlying disaggregate choice model should be developed and evaluated with particular car in the grouping of individuals with structurally different choice sets; and (f) incremental prediction should be used for prediction of the expected impacts of policy changes whever an existing set of choice shares is available to use as a basis for adjustment. /Author/

Media Info

  • Media Type: Print
  • Features: Figures; References;
  • Pagination: pp 19-24
  • Monograph Title: Passenger travel demand forecasting
  • Serial:

Subject/Index Terms

Filing Info

  • Accession Number: 00157824
  • Record Type: Publication
  • ISBN: 0309025850
  • Files: TRIS, TRB
  • Created Date: Sep 28 1977 12:00AM