01559057

Component

This study is an empirical contribution to the literature on the subject of the effect of estimation- and application- choice sets on the performance of route-choice models for short- distance trips. Route choice models were estimated using six different choice sets varying in size and composition. The models estimated with 10 or 15 alternatives with a maximum commonly factor threshold of 1 were shown to be most preferred. In the application context, the predictive performance of the models were found to decline with increasing number of alternatives in the choice set. In the current empirical context, an application choice-set size of 5 was found to yield the best predictive performance when the best models were used. Overall, these results are encouraging from a practical standpoint as models with computationally simpler choice sets are shown to have good performance over those needing additional computational costs.

This paper was sponsored by TRB committee ADB40 Transportation Demand Forecasting.

01550057

15-5880

English

22p

2015

Transportation Research Board 94th Annual Meeting

Digital/other

References; Tables

Choice models; Data files; Empirical methods; Forecasting; Mathematical prediction; Route choice; Sets

Data and Information Technology; Passenger Transportation; Planning and Forecasting; I72: Traffic and Transport Planning

Transportation Research Board Annual Meeting 2015 Paper #15-5880

TRIS, TRB, ATRI

Dec 30 2014 1:58PM