Tests of a Linear Learning Model of Destination Choice: Applications to Shopping Travel by Heterogeneous Population Groups

Abstract

It is of both theoretical and practical importance to develop improved models of the distribution of trips by a population group from given origins to given destinations for a recurrent activity. Learning models and hypotheses abound concerning the process of destination choice, but they have not yet been both mathematically specified and tested. In addition, there has been no attempt to handle the problem of developing testable statements about heterogeneous population aggregates from assumptions about individual behavior. This paper uses Bayesian probability theory to derive predictions about the successive destination choices of a group of different persons, following the specification of an appropriate linear learning model of an individual. The model's predictions are successfully tested using travel diary data on groceries' shopping by ten household groups in Uppsala. Sweden.