Date of Award

May 2018

Degree Type


Degree Name

Doctor of Philosophy



First Advisor

Jie Yu

Committee Members

Yue Liu, Xiao Qin, Matthew Petering, Changshan Wu


Data mining, Feeder bus network design, Public transit


Urbanization is not a modern phenomenon. However, it is worthwhile to note that the world urban population growth curve has up till recently followed a quadratic-hyperbolic pattern (Korotayey and Khaltourina, 2006). As cities become larger and their population expand, large and growing metropolises have to face the enormous traffic demand. To alleviate the increasing traffic congestion, public transit has been considered as the ideal solution to such troubles and problems restricting urban development. The metro is a type of efficient, dependable and high-capacity public transport adapted in metropolises worldwide. At the same time, the residents from crowded cities migrated to the suburban since 1950s. Such sub-urbanization brings more decentralized travel demands and has challenged to the public transit system. Even the metro lines are extended from inner city to outer city, the commuters living in suburban still have difficulty to get to the rail station due to the limited transportation resources.

It is becoming inevitable to develop the regional transit network such as feeder bus that picks up the passengers from various locations and transfer them to the metro stations or transportation hubs. The feeder bus will greatly improve the efficiency of metro stations whose service area in the suburban area is usually limited. Therefore, how to develop a well-integrated feeder system is becoming an important task to planners and engineers.

Realizing the above critical issues, the dissertation focus on the feeder bus network design problem (FBNDP) and contributes to three main parts:

1. Develop a data-mining strategy to retrieve OD pair from the large scale of the cellphone data. The OD pairs are able to present the users’ daily behaver including the location of residence, workplace with the timestamp of each trip. The spatial distribution of urban rail transit user demand from the OD pair will help to support the establishment and optimization of the feeder bus network. The dissertation details the procedure of data acquisition and utilization. The machine leaning is applied to predict the travel demand in the future.

2. Present a mathematical model to design the appropriate service area and routing plans for a flexible feeder transit. The proposed model features in utilizing the real-world data input and simultaneously selecting bus stops and designing the route from those targeted stops to urban rail stops.

3. Propose an improved feeder bus network design model to provide precise service to the commuters. Considering the commuters are time-sensitive during the peak hours, the time-windows of each demand is taken in to account when generating the routes and the schedule of feeder bus system. The model aims to pick up the demand within the time-windows of the commuters’ departure time and drop off them within the reasonable time. The commuters will benefit from the shorter waiting time, shorter walking distance and efficient transfer timetable.