Travel behavior analysis using smart card data

Electronic fare payment systems have gained much popularity in large public transport systems across the world. Though their main purpose is to collect the revenue and improve passenger's comfort, they also generate huge amount of detailed and accurate data on onboard transactions which is of extreme importance for transit planners for efficient short-term and long-term transit operation and service planning. This paper attempts to utilize the smart card data as an input for large-scale activity based public transport simulation for analyzing travel behavior pertaining to transit users, using an open source agent-based transport simulation package, MATSim. Transit vehicles are optimized based on input demand generated by smart card and a high temporal resolution of waiting time and its components at different stations is extracted which is very important in managing the frequency and headway of transit vehicles. Since no card swipe is needed for transfers within subway system in Seoul, the data is not readily available to detect frequent transfer locations. However, after validating the boarding and alighting volumes at different stations, a methodology is presented to calculate within subway transfer volumes at different stations. We suggest that by improving the statistical analysis and utilizing advanced data mining techniques, smart cards can effectively generate the microsimulation travel demand models.