TRB Pubsindex
Text Size:
x

« Prev Next »

Return to Results | New Search

Bookmark and Share

Title:

Comparison of Fully Latent and Partially Observed Choice Set Models for Mode Choice

Accession Number:

01764100

Record Type:

Component

Abstract:

Contemporary probabilistic choice set models assume the choice sets to be fully latent due to analyst’s lack of knowledge. While the choice set is not completely observed by the analyst, it might still be “partially observed”. Careful treatment of alternatives in the choice set based on empirical indicators could aid in completely excluding, partially/probabilistically including, and more importantly completely including some alternatives in the choice set. The paper attempts to benchmark the performance of such a partially observed choice set model with the conventional fully latent choice set models developed using empirical data collected from workers in Chennai city. The results showed that the partially observed choice set models provide a better fit with observed data than the fully latent choice set models. The consideration probability estimates from the fully latent choice set models were significantly different from 0 (1) for the completely excluded (included) alternatives in the choice set. The predicted mode shares in sub-segments also deviated considerably from the observed shares and those predicted by the partially observed choice set model. The fully latent choice set model exhibited bias in magnitudes, and the significance of some variables which can lead to misleading behavioral interpretations and erroneous policy evaluations. Thus, the additional information in the partially observed choice set model is able to more realistically evaluate the impact of policies at the consideration versus choice stages. Implications of model results for increasing the consideration and choice of public transport modes and their evaluation is illustrated under selected scenarios.

Supplemental Notes:

This paper was sponsored by TRB committee AEP50 Standing Committee on Transportation Demand Forecasting.

Report/Paper Numbers:

TRBAM-21-02977

Language:

English

Corporate Authors:

Transportation Research Board

Authors:

Kunhikrishnan, Parthan
Srinivasan, Karthik K

Pagination:

16p

Publication Date:

2021

Conference:

Transportation Research Board 100th Annual Meeting

Location: Washington DC, United States
Date: 2021-1-5 to 2021-1-29
Sponsors: Transportation Research Board; Transportation Research Board

Media Type:

Digital/other

Features:

Figures; References; Tables

Geographic Terms:

Chennai (India)

Subject Areas:

Planning and Forecasting; Policy; Public Transportation

Source Data:

Transportation Research Board Annual Meeting 2021 Paper #TRBAM-21-02977

Files:

TRIS, TRB, ATRI

Created Date:

Dec 23 2020 11:19AM