Choice modeling with observed and unobserved information search
Author(s)Xie, Yifei,S.M.Massachusetts Institute of Technology.
Massachusetts Institute of Technology. Department of Civil and Environmental Engineering.
Moshe E. Ben-Akiva.
MetadataShow full item record
This thesis contributes to the continuous effort of enhancing discrete choice models with richer behavior representations by explicitly modeling the information search process leading up to a choice. Information search includes information acquisition as well as information processing, the respective observability of which often raises challenges in the modeling. The observability of information acquisition is contingent on the context of the choice situation and data collection, while information processing is a mental process that is by nature latent to the modelers. First part of the thesis considers the specific case of information search under the context of smart mobility services. Leveraged on clickstream data, the information acquisition in this case is observable. With the assumption that the acquired information would be fully processed, a modeling framework is proposed to directly model the observed information search decision, and integrate it with other relevant decisions in smart mobility usage including subscription, menu choice and opt-out choice. The framework is illustrated through Tripod, a mobility service which provides on-demand incentives for sustainable travel behaviors. The second part of the thesis considers the case where the decision-maker might not process all the acquired information. A search action model is proposed to fully acknowledge the latency of the information processing behavior and hence account for its influence on the final choice. This model assumes that the decision-maker has acquired the information and only considers the information processing behavior. As an extension to standard random utility maximization (RUM) models, the framework is especially relevant for modern travel behavior modeling where real-time travel information is often readily available but not always considered in decision-making. A preliminary Monte Carlo experiment is conducted to validate model identification and estimation.
Thesis: S.M. in Transportation, Massachusetts Institute of Technology, Department of Civil and Environmental Engineering, 2019Cataloged from PDF version of thesis.Includes bibliographical references (pages 73-80).
DepartmentMassachusetts Institute of Technology. Department of Civil and Environmental Engineering
Massachusetts Institute of Technology
Civil and Environmental Engineering.