Effects of Less-Equilibrated Data on Travel Choice Model Estimation

Oh, Jun-Seok; Cortés, Cristián; Recker, Will

Abstract

Most discrete choice models assume steady state conditions and a fully equilibrated system when estimating unknown coefficients from real-world data. However, the estimated model can be biased when the data set used for the model estimation was drawn from non- or less-equilibrated traveler behavior. The resulting biased model could lead to a misunderstanding of the system. Such effects on discrete choice model estimation were examined by performing Monte Carlo simulation experiments. A day-to-day dynamic evolutionary framework was used to observe changes in traveler's choice and to compare the estimated results during the adjustment process with the true behavior parameters.

Más información

Título de la Revista: TRANSPORTATION RESEARCH RECORD
Volumen: 1831
Editorial: SAGE PUBLICATIONS INC
Fecha de publicación: 2003
Página de inicio: 131
Página final: 140
DOI:

10.3141/1831-15

Notas: ISI