ABSTRACT
A B S T R A C T The disaggregate nature of transit smart card data is congruent with the travel demand specification as used by agent-based approaches to transport modelling. Using a full day of public transport smart card transactions recorded in Singapore, we develop an approach to transform the smart card data into both transport supply and demand, while simultaneously eliminating the need to simulate the interaction between cars and buses. In order to produce realistic travel times for buses, a regression model of bus speed between stops was estimated, that is dependent both on the level of demand and network topology. The simulation includes a model of bus dwell time at stops that is dependent on the ridership of the bus and its configuration. It allows us to simplify the supply network dramatically with only one link between bus stop combinations and another link at the stop for buses to queue to perform dwell operations. These modifications, along with a simplified mobility simulation, dramatically improves simulation times, ensuring useable results in an hour. In addition, our modelling framework is highly adaptable and requires only limited efforts to be applied to other public transport systems in cities where similar data streams are available.
