Use of concepts of regret and disappointment to model dynamic route choice behavior under uncertainty

This paper reports progress in developing a dynamic disappointment- and regret-based route choice model that incorporates the learning of uncertain travel times. Numerical simulations were conducted to measure the effects of varying travel time distributions of available routes, parameters for sensitivity to regret and disappointment, and initial guesses that travelers have for the routes’ travel times. Four simulation scenarios were explored. Of the varied inputs, the interaction between the travel time distribution of the routes and the magnitudes of the parameters for regret and disappointment was what affected shares more strongly. It was found that disappointment-averse travelers preferred routes with lower variability, even if the average travel times were higher. Regret-averse travelers, however, took both the mean and variance of the routes into consideration for making decisions and preferred routes with higher chances of generating lower outcomes, even if their variabilities were high. For the number of iterations adopted, no direct effect was found from varying the initial priors on the shares, although narrow distributions for the initial guesses may strongly affect learning and, therefore, indirectly affect shares.