Special issue: Public transit and pedestrian studies

Abstract

In the past decade, public transportation and pedestrian traffic issues have gained rapidly increasing attention in both academia and industry, partly thanks to the greater appreciation of sustainable urban transport. Recent advances in data analytics have further motivated research on better estimation methods and management strategies in the realms of public transit and pedestrian studies.One prevailing topic in these realms is transit service reliability, which has long been recognized as a dominant factor of service quality and ridership of transit systems [1,2]. To explore the causes of unreliability, better models of transit travel time variability are needed [3]. Another topic of broad interest is passenger demand estimation, which proved to be essential for both short-term and longterm service planning [4]. This includes the estimation of passenger arrival, departure, and transfer flows at stations of a complex transit network [5,6]. The estimation results can be further used, for example, to jointly optimize passenger travel time and fuel consumption for rail transit [7].On the pedestrian side, strategies for evacuating occupants from a facility have been the focus of a number of studies (e.g., [8]). Another topic of wide concern is the designs for improving walkability, for example, via human-centered street designs [9].This special issue, including six select papers, was well suited to the emerging research needs described in the preceding text. Four out of the six papers are in the realm of transit, which presented more advanced models and approaches (e.g., mixture models, Box-Jenkins method, and multi-class traffic assignment) for improving the accuracy, efficiency, and automatization of demand forecast, travel time estimation, and service optimization of real-world transit systems. The other two are pedestrian studies, which pertain to the optimal ways for pedestrian evacuation through a complex queueing network, and the perceptions of human-centered street design elements, respectively. More details are furnished as follows.We start from the work by Ma et al.[10] on modeling the distributions of day-to-day bus travel times, which is a key prerequisite for transit reliability analysis. Although there have been ample empirical studies in this realm, those studies unveiled inconsistent conclusions because of the limitation of data sets and evaluation approaches. To this end, the authors proposed a novel evaluation approach and a more complete set of performance measures (including not only the fitting accuracy, but also the robustness and the explanatory power) to identify better travel time-distribution models in the category of mixture models. Using automatic vehicle location data collected on two typical bus routes in Brisbane over a period of 6 months, the Gaussian Mixture Models were shown to be superior to its alternatives (e.g., normal, Weibull, lognormal, and log-logistic distributions, which have been widely used in the literature). The analysis also showed that the levels of spatial and tem...