ORIGIN AND DESTINATION INFERENCE OF BUS PASSENGERS
Ilgin Gokasar, Bogazici University, Akif Fidanoglu, Bogazici University
A methodology is proposed to infer destinations of passengers using ADC information of public transportation systems and to generate the OD matrices for bus passengers using a case study for a single bus route in İstanbul.
With the advance technology used in the public transportation systems it has become much easier to monitor the trips of the passengers. Most of the public transportation systems record the passengers when they start their trip with the help of Automated Data Collection (ADC) systems installed to the vehicles. By this way, transit agencies records the origins of the passengers’ trips. However, it is a bit challenging issue to detect the passengers’ destination, especially for the bus passengers since there is no data recorded when the passengers exit from the system. This study explores the methods to infer the destination of the passengers and to generate the Origin-Destination (OD) matrices for the bus passengers by conducting a case study for a single bus route in İstanbul.
To estimate the destinations of the passengers, several assumptions have been set in the previous studies. In this study, these assumptions were used to infer the alighting location of the bus passengers with further assumptions based mainly on the characteristics of transportation system in Istanbul. Proposed algorithm for alighting inference of bus passengers is changed with additional steps. For the interchanges in short time interval, interchange trip information is assumed to be more reliable than the recorded boarding location and direction of the bus for previous trip. Also to detect the repetitive use of the same smartcard by different passengers for the same trip, usage of the same smartcard in close time interval and in the same bus is examined in detailed. For the last trips of the day, in addition to the methods used in previous studies, another method which uses the recorded next trip of the passenger on upcoming days to infer the alighting location, is introduced.
Even though there were huge amount of missing records in acquired data for boarding locations of the passengers, with the help of additional assumptions in the proposed method, for 65 % of the studied trip alighting location of the bus passengers could be successfully inferred. Validation is made using the comparison of the inferred results and the surveys made on the same route.
Association for European Transport