Estimation of the Cross-Nested Logit Model for a Large Number of Alternatives

Most activity-based models are composed by a series of Nested Logit (NL) models with a fixed hierarchical structure. The Cross-Nested Logit (CNL) model is a generalization of the NL model, since it allows an alternative to belong to more than one nest, and thus allowing the specification of more flexible structures. The flexibility of the CNL model comes at the expense of estimating additional parameters that allocate the alternatives to nests. In transportation choice problems with a large number of alternatives, the number of allocation parameters to estimate may increase rapidly. The motivation of this paper is to find practical methods to infer the CNL allocation parameters for large size problems, such as a system of combined decisions. The paper illustrates the problem by estimating a combined mode and destination choice model for shopping trips using revealed preference data. A simple method to define the allocation parameters is suggested, and the correspondent CNL model outperforms other CNL and NL specifications.

Language

  • English

Media Info

  • Media Type: CD-ROM
  • Features: Figures; References; Tables;
  • Pagination: 15p
  • Monograph Title: TRB 86th Annual Meeting Compendium of Papers CD-ROM

Subject/Index Terms

Filing Info

  • Accession Number: 01047328
  • Record Type: Publication
  • Report/Paper Numbers: 07-0662
  • Files: TRIS, TRB
  • Created Date: Feb 8 2007 5:15PM