|345-"运输与物流讲堂”第九讲-Travel behavior analysis with passive data stream: Case studies using UK and Japan smart Card data|
主讲人：Dr. Amr M. Wahaballa
Dr. Amr M. Wahaballa is an Associate Professor at Aswan University, Egypt and a researcher at Gifu University, Japan. He received his BSc (Civil Eng) and MSc (Transportation Eng) degrees from South Valley University of Egypt and a PhD in Transportation Engineering from Gifu University of Japan. Dr. Wahaballa has published extensively in reputable international journals with high impact factors in addition to numerous conference papers and presentations. He is currently doing research in Japan for 6 years ago with Gifu, Nagoya and Kyoto Universities. He is a member of public Transport committee of the Transportation Research Board and is a registered reviewer for Transportation Research Record, International Journal of Intelligent Transportation Systems Research, Transportation Research Part C, Journal of Intelligent Transportation Systems, and other related journals including IEEE Transactions on Intelligent Transportation Systems, etc.
Urban transit networks serve many passengers and play an important role in urban mobility. In dense transit networks, operators need to understand travel patterns to preserve a good service level. However, passenger route choice behavior is not directly observable and is affected by many factors such as in-vehicle time, transfer time, waiting time and their variabilities. Researchers made different assumptions on these factors and deduced variable results which affect the transit management accuracy. Therefore, transit travel time estimation has so far received little attention in the research literature. This research aims to estimate travel time distributions on the London underground and Shizuoka bus and rail networks considering travel time variability by using smart card data. The two different datasets are rich with practical information that lead to promising results that can be handled for use in real-world practice. The stochastic frontier model is used, and its parameters are estimated by the maximum likelihood method. The cost frontier function is used to represent the relation between the travel time (observed by the smart card) as an output and the in-vehicle time and walking time (from supplementary data) as inputs. Comparing the travel time values estimated by the proposed model with the values observed by the smart card shows a high goodness-of-fit. The estimation proves to be a promising quick convergence and computationally efficient. The results could facilitate improvements in the transit service reliability analysis and the passenger flow assignment. Matching the obtained distributions with smart card observations will help with estimating route choice behavior that can validate current transit assignment models.