Modeling learning in route choice Bogers, Enide AI ; Bierlaire, Michel ; Hoogendoorn, Serge P
Series: ; 2014Publication details: Transportation research record, 2007Description: s. 1-8Subject(s): Bibl.nr: VTI P8167:2014Location: Abstract: Performing the same trip many times, travelers can learn about available routes from their experiences. Two types of learning found in psychological learning theory appear to play a role in day-to-day route choice: implicit (reinforcement-based) and explicit (belief-based). Memory decay also plays a major role. Although much progress had been made in modeling learning in route choice, a model that captures both learning types and for which the parameters are empirically underpinned was not found. Such a model thus is developed, and a large data set from experimental research is used to validate it and to estimate its parameters. The developed model uses a Markov formulation for the day-to-day updating of a person's belief about travel time (i.e., perceived travel time) on a route. Reinforcement (and inertia) is modeled by including the latest 10 route choices in the model. Results indicate that 20% of perceived travel time is from the most recent experience; therefore, formulations that use either the mathematical mean of all past experienced travel times or only the most recent travel times are not accurate. Furthermore, the reinforcement-inertia part of the model can make up a significant part of the route utility and therefore should be a standard component in route choice models. In sum, the results validate the theoretical and mathematical model.Current library | Status | |
---|---|---|
Statens väg- och transportforskningsinstitut | Available |
Performing the same trip many times, travelers can learn about available routes from their experiences. Two types of learning found in psychological learning theory appear to play a role in day-to-day route choice: implicit (reinforcement-based) and explicit (belief-based). Memory decay also plays a major role. Although much progress had been made in modeling learning in route choice, a model that captures both learning types and for which the parameters are empirically underpinned was not found. Such a model thus is developed, and a large data set from experimental research is used to validate it and to estimate its parameters. The developed model uses a Markov formulation for the day-to-day updating of a person's belief about travel time (i.e., perceived travel time) on a route. Reinforcement (and inertia) is modeled by including the latest 10 route choices in the model. Results indicate that 20% of perceived travel time is from the most recent experience; therefore, formulations that use either the mathematical mean of all past experienced travel times or only the most recent travel times are not accurate. Furthermore, the reinforcement-inertia part of the model can make up a significant part of the route utility and therefore should be a standard component in route choice models. In sum, the results validate the theoretical and mathematical model.