Transaction data from public transit smart cards represent a continuous stream of detailed travel information for transit demand modeling. Although certain aspects of information are incomplete in unprocessed data, efforts are devoted to deriving a more comprehensive understanding of the system and its users from partial information through data enrichment processes, with a long-term goal of establishing a dynamic model of demand. On the basis of previous work, methods are proposed to estimate the arrival time of bus runs at the stop level by using temporal constraints and to identify linked trips by using spatial–temporal concepts. These enrichments lead to the reconstruction of individual itineraries, the analysis of transfer activity, and the synthesis of vehicle load profiles. The latter provide planners with a detailed spatial–temporal progression of each run, origin and destination stops for each individual transaction, and boarding and alighting activity at each stop. The study draws on more than 37,000 smart card boarding transactions of an average weekday from a midsize transit agency. Results suggest that linked trips represent slightly above 10% of the total number of transactions in the network and the smart card system overestimates the proportion of linked trips by nearly 40%. The outcome is promising and lays a foundation to further enrich the itineraries by associating the boarding and alighting stops with trip generators, deriving trip purposes, and performing multiday analysis.

Trips need to be described and have always been characterized by various levels of abstraction. It varies from a simple label such as home-based work to complete itinerary with sociodemographic characteristics of the trip maker and household. The rationale behind such classifications is that planners and modelers recognize that the demand of transportation is highly differentiated. It is hoped that additional attributes would provide a more complete portrait of the demand and an improved understanding of the underlying travel behavior. Passive data collection technologies bring an extra dimension to travel data acquisition. Multiday data, which are difficult to collect, become accessible. In public transit, a smart card automatic fare collection system with automatic vehicle location capability provides high-resolution longitudinal data on travel pattern but also suffers from the inherent limitations of passive methods. This paper proposes a methodology to enhance transit trip characterization by adding a multiday dimension to a month of smart card transactions. On the basis of an individual, anchor points—precise to an exact address—are detected. Boarding and alighting locations are described with respect to those anchors. The enhancement allows in-depth travel behavior analysis on a subgroup sharing a common anchor or an individual. The paper demonstrates the use of spatial statistics, spatial analyses with geographic information system, visualizations, and data mining to describe activity space and locations and departure time dynamics, and to derive monthly trip table, activity schedule, and behavioral rules for cardholders. The results offer promising insights to transit planning and the understanding of travel behavior.

No abstract