Abstract
Emerging transport modeling approaches such as agent/activity-based modeling as well as the shift toward data-driven paradigm have stressed the need for high-quality travel behavior data at a disaggregate level. Despite the advances in personal mobility tracking and capturing, aggregate methods such as origin–destination (OD) matrices are still the most widespread means to organize and represent travel demand information. Nonetheless, traditional ODs cannot directly capture substantial elements which can considerably affect travel behavior (e.g., trip interdependency, trip chaining, etc.) therefore they are not the most suitable mean to facilitate relevant analysis. The currently presented framework alleviates this limitation by combining the individual trips present in multi-period, purpose-dependent OD matrices into sequences of trips originating and ending at users’ home location (i.e., tours). This is achieved by integrating graph-theoretical with combinatorial optimization concepts. A graph theory-based methodology is applied to first examine the spatiotemporal information in ODs and then identify all the plausible tours. Then the sequence of activities (i.e., daily activity schedule [DAS]) taking place during each diurnal tour is inferred based on the trip-purpose information contained in the original ODs. Finally, a combinatorial optimization routine identifies the optimum combination of DAS whose aggregated trips represent the total travel demand as observed in the OD matrices. The resulting DAS by the proposed estimation framework has proven to be of significant accuracy and provide valuable information in relation to the characteristics of the population’s travel behavior within the urban space environment. This information is particularly useful to mitigate against the upcoming challenges in the field of smart mobility management.
Get full access to this article
View all access options for this article.