Exploiting the relation between Activity Data and Traffic Data within the Dynamic Demand Estimation Problem

The Dynamic Demand Estimation (DDE) problem searches for time-dependent demand matrices; to solve this problem, in the classical approaches, survey data and traffic data are used [1-4]. In this paper an explorative analysis is conducted, with the goal to capture and analyze the existing relation between the traffic demand observed with the classical traffic counts and the activity patterns of the users. The aim is to exploit information from daily and weekly activity-travel patterns to use as input in the DDE problem, for example to enrich the (time-dependent) seed matrix. Activity data are already studied in the traffic engineering for different applications; many works in the literature underline the relevance to establish a relation between the trip chains and the activities within the day for each user, highlighting that generally the trip is motivated from a specific purpose (Ettema and Timmermans [5]). A fundamental aspect in the research is to understand how the users of the traffic network schedule their trips taking into account different elements related to recurrent and non-recurrent household activities [6]. Various scheduling models are proposed in literature to obtain the scheduling of the activity pattern and activity travel behavior ([7-8]) and integrate them in travel choice processes using a supernetwork approach [9]. These models are directly combined in simulation tools, like Albatross (TU/e [8]) and MatSIM (ETH, TU Berlin [10, 11]). In this case the traffic simulation takes into account the activity scheduling of the users to estimate the departing time. The research on the activity-based models is becoming even more important as more and more activity-related data are available through the new information and communication technologies and the spread of social networks; data collected from social platforms such as “foursquare” provide frequent, accurate and cheap information on various activities, their location and the check-in times of social network users. Similar features can be found in other social networks (e.g. Facebook, twitter). On the other hand it is to highlight that, although the authors stress the relation between departure time and activity information, there are only few works that use this information in the demand estimation problem and generally they are limited to the static Demand Estimation problem; A recent example in this case is a case study on the city of Austin [12].