The prediction of passenger demand in public transport by using smart card data and scheduling optimization

In the countries with limited economic sources, public transportation systems are based on bus transit rather than rail transit systems because rail transit systems have higher initial investment costs. Although bus transit systems can provide flexibility of bus line planning, operating and revision, they are mostly operated on mixed vehicle traffic and thus their performance directly is affected by the traffic conditions. Due to the traffic conditions delays increase especially in the peak hours and the service quality of bus transit system decreases considerably. Smart card based fare collection systems are widely used by transit operators. While their main purpose is perceived as collecting revenue, smart cards provide large quantities of detailed data. Evaluation of obtained data provides useful results for improving short and long term operational strategies.In the first stage of the thesis, facilities of transit systems in İzmir city center and passenger tendencies are discussed. By 2012, public transportation service is provided by light rail system, commuter rail system, ferry routes, bus routes and para-transit routes. Except the para-transit routes, other types are operated by İzmir Metropolitan Municipality. When smart card data are evaluated, it is seen that the 80% proportion of all boarding in a week day is supplied by bus routes. This clearly shows that highway based transportation systems are used widely. Smart card data are evaluated with generated algorithms and passenger, hour, route, bus stop based results were obtained. By using these results, it is obtained that the routes giving services between city center and high density residence areas, offering transfers with other transit systems and the routes providing transportation to the university campuses have higher passenger loads. Besides, most of the bus stops that have higher passenger loads are determined to have limited waiting areas. It is also seen that, especially transfer centers that provides integration between bus routes and commuter rail or ferry routes have lower passenger loads which show that the integration between the transit systems are not efficiently provided.One of the problem that tried to be solved in the thesis is, determining the optimum trip frequencies that take into account the passenger demand daily changes. In the study, optimum frequencies depending on passenger demand are investigated for the sample bus routes which serve between South-East region of İzmir city and city center Konak. Optimum frequencies that can be used for morning peak period are calculated by using some empirical equations and linear goal programming methods and results are compared with operator's.