A cumulative prospect theory approach to commuters' day-to-day route-choice modeling with friends' travel information. (January 2018)

• Record Type: Journal Article
• Authors: Zhang, Chong; Liu, Tian-Liang; Huang, Hai-Jun; Chen, Jian
• Is Part Of: Transportation research. Volume 86(2018)
• Page Start: 527
• Page End: 548
• Publication Date: 2018-01
• Subjects: Route choice -- Social interaction -- Day-to-day dynamics -- Cumulative prospect theory; Transportation -- Periodicals; Transportation -- Technological innovations -- Periodicals; 388.011
• DOI: 10.1016/j.trc.2017.12.005
• Languages: English
• ISSNs: 0968-090X

Abstract

Highlights: A day-to-day route-choice model with friends' information is developed. A behavioral experiment is conducted to collect thedata associated with subjects' choice decisions. A larger rate of social interactions does not necessarily lead to a better outcome. Forecast accuracies with and without friends' information are compared. The model is improved by incorporating the overlapping effects of routes. Abstract: This paper investigates the effects of social interaction information from friends on commuters' daily route choice decisions. Besides the actual route travel time shared among friends, both the amount and percentage of friends choosing each route are regarded as being influence factors. For estimating the factors' relative importance, this paper first develops a day-to-day route-choice learning model with friends' travel information based on the Cumulative Prospect Theory (CPT), and then designs and conducts a laboratory behavioral experiment to collect the statistical data associated with subjects' actual route choice decisions. Experimental results show that a larger rate of social interactions in an online travel community does not necessarily lead to a better route-choice outcome for individuals or the whole system. Furthermore, the overall impact of the amount and percentage of friends choosing each route on the generation of perceived travel time may be negative or positive, depending on the number of members in an online travel community. Using the developed model, the endogenous reference points of the subjects are estimated to first increase and then decrease over simulated days till being roughly leveling off, and the average travel prospect values of the subjects on all routes are estimated to first increase, then decrease and finally level off over simulated days. We also discuss the implication of integrating friends' travel information into modeling by comparing the forecast accuracies of the models with and without direct consideration of friends' travel information, and improve the developed model by incorporating the overlapping effects of routes.