Non-Unimodal Nested Logit Models with Non-Linear Utility Functions

This paper deals with the already classical problem of modal split estimation based on discrete choice data within the framework of nested logit models. The formulation is based on the definition of non-linear utility functions through the Box-Cox transform. The outcome of this research proves the non-unimodal character of the objective function defined by the log-likelihood estimation method. The existence of multiple local optima with close values of the objective for disperse estimators permits the choice of the most adequate one in terms of additional requirements such as robustness of the inferred model and sensitivity to utility attribute variations.

• Corporate Authors:

  World Conference on Transport Research Society

  Secretariat, 14 Avenue Berthelot
  69363 Lyon cedex 07,   France 
• Authors:
  • Benitez, Francisco G
  • Vazquez, Javier
• Conference:
  • 11th World Conference on Transport Research
  • Location: Berkeley CA, United States
  • Date: 2007-6-24 to 2007-6-28
• Publication Date: 2007

Language

• English

Media Info

• Media Type: CD-ROM
• Features: Figures; References; Tables;
• Pagination: 29p
• Monograph Title: 11th World Conference on Transport Research

Subject/Index Terms

• TRT Terms: Public transit; Sensitivity analysis; Transit operating agencies; Transportation planning
• Uncontrolled Terms: Discrete choice; Nested logit models
• Subject Areas: Highways; Planning and Forecasting; Public Transportation; I72: Traffic and Transport Planning;

Filing Info

• Accession Number: 01118444
• Record Type: Publication
• Files: TRIS
• Created Date: Jan 13 2009 9:30AM