Modelling activity conduction preferences and their effect on departure time choicesRevista : Transportation Research Part A-Policy and Practice
Volumen : 129
Páginas : 39-51
Tipo de publicación : ISI Ir a publicación
A great challenge that comes with cities growth and urbanization is how to transport their inhabitants efficiently. Since available urban space has been progressively reducing, the attention has shifted from capacity investment to transport demand management. As a result, policies aimed at changing departure time choices have arisen. To properly analyze such policies, it is necessary to understand how and when people prefer to travel, which is a result of the activities individuals perform. This paper studies preferences towards activities to be conducted and the impact of these preferences on departure time choices. For this purpose, a stated preference survey was developed to reproduce a daily activity pattern for all respondents. With the collected data it was possible to compare alternative approaches to model activities proposed in the literature. This comparison shows that traditional formulations to model activities preferences might not be the ideal in some cases. Instead, the use of more complex formulations, such as sigmoidal functions, is encouraged since they could adequately represent the preferences regarding activity’s duration and time of day, as well as activity’s flexibility. Results show that activities to be conducted before and after each trip affect the departure time choices and should be considered when modelling and forecasting travel demand. Therefore, a proper benefit estimation from travel time savings should include a direct and indirect component: one related to avoid travelling (which is an undesired activity by itself) and another one related to the increase of available time to perform activities instead of travelling. Our results show that if the latter effect is not included, the value of mean travel time would be underestimated by 38% and total benefits could be miscalculated.