A robust method for estimating transit passenger trajectories using automated data

Development of an origin-destination demand matrix is crucial for transit planning. The development process is facilitated by automated transit smart card data, making it possible to mine boarding and alighting patterns on an individual basis. This research proposes a novel trip chaining method which uses Automatic Fare Collection (AFC) and General Transit Feed Specification (GTFS) data to infer the most likely trajectory of individual transit passengers. The method relaxes the assumptions on various parameters used in the existing trip chaining algorithms such as transfer walking distance threshold, buffer distance for selecting the boarding location, time window for selecting the vehicle trip, etc. The method also resolves issues related to errors in GPS location recorded by AFC systems or selection of incorrect sub-route from GTFS data. The proposed trip chaining method generates a set of candidate trajectories for each AFC tag to reach the next tag, calculates the probability of each trajectory, and selects the most likely trajectory to infer the boarding and alighting stops. The method is applied to transit data from the Twin Cities, MN, which has an open transit system where passengers tap smart cards only once when boarding (or when alighting on pay-exit buses). Based on the consecutive tags of the passenger, the proposed algorithm is also modified for pay-exit cases. The method is compared to previous methods developed by the researchers and shows improvement in the number of inferred cases. Finally, results are visualized to understand the route ridership and geographical pattern of trips.     

Calibrating a transit assignment model using smart card data in a large-scale multi-modal transit network

This paper describes a practical automated procedure to calibrate and validate a transit assignment model. An optimization method based on particle swarm algorithm is adopted to minimize a defined error term. This error term which is based on the percentage of root mean square error and the mean absolute percent error encompasses deviation of model outputs from observations considering both segment level as well as the mode level and can be applied to a large scale network. This study is based on the frequency-based assignment model using the concept of optimal strategy while any transit assignment model can be used in the proposed methodological framework. Lastly, the model is validated using another weekday data. The proposed methodology uses automatic fare collection (AFC) data to estimate the origin–destination matrix. This study combines data from three sources: the general transit feed specification, AFC, and a strategic transport model from a large-scale multimodal public transport network. The South-East Queensland (SEQ) network in Australia is used as a case study. The AFC system in SEQ has voluminous and high quality data on passenger boardings and alightings across bus, rail and ferry modes. The results indicate that the proposed procedure can successfully develop a multi-modal transit assignment model at a large scale. Higher dispersions are seen for the bus mode, in contrast to rail and ferry modes. Furthermore, a comparison is made between the strategies used by passengers and the generated strategies by the model between each origin and destination to get more insights about the detailed behaviour of the model. Overall, the analysis indicates that the AFC data is a valuable and rich source in calibrating and validating a transit assignment model.

     

Estimation of population origin–interchange–destination flows on multimodal transit networks

Previous research has combined automated fare-collection (AFC) and automated vehicle-location (AVL) data to infer the times and locations of passenger origins, interchanges (transfers), and destinations on multimodal transit networks. The resultant origin–interchange–destination flows (and the origin–destination (OD) matrices that comprise those flows), however, represent only a sample of total ridership, as they contain only those journeys made using the AFC payment method that have been successfully recorded or inferred. This paper presents a method for scaling passenger-journey flows (i.e., linked-trip flows) using additional information from passenger counts at each station gate and bus farebox, thereby estimating the flows of non-AFC passengers and of AFC passengers whose journeys were not successfully inferred.The proposed method is applied to a hypothetical test network and to AFC and AVL data from London’s multimodal public transit network. Because London requires AFC transactions upon both entry and exit for rail trips, a rail-only OD matrix is extracted from the estimated multimodal linked-trip flows, and is compared to a rail OD matrix generated using the iterative proportional fitting method.     

Development of a transit network from a street map database with spatial analysis and dynamic segmentation

This paper presents an integrated transit-oriented travel demand modeling procedure within the framework of geographic information systems (GIS). Focusing on transit network development, this paper presents both the procedure and algorithm for automatically generating both link and line data for transit demand modeling from the conventional street network data using spatial analysis and dynamic segmentation. For this purpose, transit stop digitizing, topology and route system building, and the conversion of route and stop data into link and line data sets are performed. Using spatial analysis, such as the functionality to search arcs nearest from a given node, the nearest stops are identified along the associated links of the transit line, while the topological relation between links and line data sets can also be computed using dynamic segmentation. The advantage of this approach is that street map databases represented by a centerline can be directly used along with the existing legacy urban transportation planning systems (UTPS) type travel modeling packages and existing GIS without incurring the additional cost of purchasing a full-blown transportation GIS package. A small test network is adopted to demonstrate the process and the results. The authors anticipate that the procedure set forth in this paper will be useful to many cities and regional transit agencies in their transit demand modeling process within the integrated GIS-based computing environment.     

A bi‐level programming approach — Optimal transit fare under line capacity constraints

The fare of a transit line is one of the important decision variables for transit network design. It has been advocated as an efficient means of coordinating the transit passenger flows and of alleviating congestion in the transit network. This paper shows how transit fare can be optimized so as to balance the passenger flow on the transit network and to reduce the overload delays of passengers at transit stops. A bi‐level programming method is developed to optimize the transit fare under line capacity constraints. The upper‐level problem seeks to minimize the total network travel time, while the lower‐level problem is a stochastic user equilibrium transit assignment model with line capacity constraints. A heuristic solution algorithm based on sensitivity analysis is proposed. Numerical example is used to illustrate the application of the proposed model and solution algorithm.     

Transit assignment under crowded conditions

This paper presents a transit assignment algorithm for crowded networks. Both congestion in vehicles and queuing at stations are explicitly taken into account in predicting passenger flows for a fixed pattern of origin-destination trip demands. The overflow effects due to insufficient capacity of transit lines are considered to be concentrated at transit stations, while the in-vehicle congestion effects (or discomforts) are considered to be dependent on in-vehicle passenger volume. Overflow delay at a transit station is dependent on the number of excess passengers required to wait for the next transit car. We use a logit model to determine the split between passengers that chose to wait for the next transit car and passengers that chose to board on the alternative transit lines. The proposed algorithm predicts how passenger will choose their optimal routes under both queuing and crowded conditions.     

Estimating the influence of common disruptions on urban rail transit networks

With the continuous expansion of urban rapid transit networks, disruptive incidents—such as station closures, train delays, and mechanical problems—have become more common, causing such problems as threats to passenger safety, delays in service, and so on. More importantly, these disruptions often have ripple effects that spread to other stations and lines. In order to provide better management and plan for emergencies, it has become important to identify such disruptions and evaluate their influence on travel times and delays. This paper proposes a novel approach to achieve these goals. It employs the tap-in and tap-out data on the distribution of passengers from smart cards collected by automated fare collection (AFC) facilities as well as past disruptions within networks. Three characteristic types of abnormal passenger flow are divided and analyzed, comprising (1) “missed” passengers who have left the system, (2) passengers who took detours, and (3) passengers who were delayed but continued their journeys. In addition, the suggested computing method, serving to estimate total delay times, was used to manage these disruptions. Finally, a real-world case study based on the Beijing metro network with the real tap-in and tap-out passenger data is presented.     

Analyzing year-to-year changes in public transport passenger behaviour using smart card data

In recent years, there has been increased interest in using completely anonymized data from smart card collection systems to better understand the behavioural habits of public transport passengers. Such an understanding can benefit urban transport planners as well as urban modelling by providing simulation models with realistic mobility patterns of transit networks. In particular, the study of temporal activities has elicited substantial interest. In this regard, a number of methods have been developed in the literature for this type of analysis, most using clustering approaches. This paper presents a two-level generative model that applies the Gaussian mixture model to regroup passengers based on their temporal habits in their public transportation usage. The strength of the proposed methodology is that it can model a continuous representation of time instead of having to employ discrete time bins. For each cluster, the approach provides typical temporal patterns that enable easy interpretation. The experiments are performed on five years of data collected by the Société de transport de l’Outaouais. The results demonstrate the efficiency of the proposed approach in identifying a reduced set of passenger clusters linked to their fare types. A five-year longitudinal analysis also shows the relative stability of public transport usage.     

Rescuing the captive [mode] user: an alternative approach to transport market segmentation

The prevailing approach to transport market segmentation which identifies two distinct groups, “captive” and “choice” users, has widely been used by professionals and scholars despite the ambiguity associated with these terms. Furthermore, conflicting interpretations from the point of view of decision makers and individuals may result in negative policy implications where the needs of captive users are neglected in favour of attracting new users. This study attempts to address these concerns by proposing an alternative segmentation framework that could be applied to any mode of transport, in any regional context, by users and decision makers alike to better guide the development of transport policies. Using the results of a large-scale transportation survey, a series of clustering techniques are employed to derive this alternative approach for segmenting walkers, cyclists, transit and automobile users. The main factors considered in the final clustering analysis are the level of trip satisfaction and practicality. The analysis yielded four market segments: captivity, utilitarianism, dedication and convenience. Using this theoretical framework to understand the distribution of travellers among market segments is essential in identifying distinct and appropriate policy interventions to improve trip conditions. It is hoped that the segmentation approach and policy framework proposed here will encourage a better balance between pragmatic and idealistic goals in transportation policy.     

Transit network design based on travel time reliability

This paper presents a transit network optimization method, in which travel time reliability on road is considered. A robust optimization model, taking into account the stochastic travel time, is formulated to satisfy the demand of passengers and provide reliable transit service. The optimization model aims to maximize the efficiency of passenger trips in the optimized transit network. Tabu search algorithm is defined and implemented to solve the problem. Then, transit network optimization method proposed in this paper is tested with two numerical examples: a simple route and a medium-size network. The results show the proposed method can effectively improve the reliability of a transit network and reduce the travel time of passengers in general.     

User perspectives in public transport timetable optimisation

The present paper deals with timetable optimisation from the perspective of minimising the waiting time experienced by passengers when transferring either to or from a bus. Due to its inherent complexity, this bi-level minimisation problem is extremely difficult to solve mathematically, since timetable optimisation is a non-linear non-convex mixed integer problem, with passenger flows defined by the route choice model, whereas the route choice model is a non-linear non-continuous mapping of the timetable. Therefore, a heuristic solution approach is developed in this paper, based on the idea of varying and optimising the offset of the bus lines. Varying the offset for a bus line impacts the waiting time passengers experience at any transfer stop on the bus line.In the bi-level timetable optimisation problem, the lower level is a transit assignment calculation yielding passengers’ route choice. This is used as weight when minimising waiting time by applying a Tabu Search algorithm to adapt the offset values for bus lines. The updated timetable then serves as input in the following transit assignment calculation. The process continues until convergence.The heuristic solution approach was applied on the large-scale public transport network in Denmark. The timetable optimisation approach yielded a yearly reduction in weighted waiting time equivalent to approximately 45 million Danish kroner (9 million USD).     

Empirical analysis and comparisons about time-allocation patterns across segments based on mode-specific preferences

A three-stage approach, i.e., factor-cluster-multi-group Structural Equation Modeling (SEM), is designed to explore the commonalities and diversities with respect to relationships between socio-demographic characteristics and time-use patterns across different segments. Factor-cluster analysis is conducted to extract meaningful factors from attitudinal statements, and then group the sample population into three segments, each with a unique combination of mode preferences for public transit, private car, and motorcycle. By virtue of multi-group SEM, the relationships between socio-demographics and time allocated to activities and travel are found to be significantly different across segments. This study highlights the importance of latent psychological factors in segmentation. For policy implication, specific population with unique psychological features must be targeted in order to efficiently and effectively design and implement transport measures.     

Large-scale application of MILATRAS: case study of the Toronto transit network

This paper documents the efforts to operationalize the conceptual framework of MIcrosimulation Learning-based Approach to TRansit Assignment (MILATRAS) and its component models of departure time and path choices. It presents a large-scale real-world application, namely the multi-modal transit network of Toronto which is operated by the Toronto Transit Commission (TTC). This large-scale network is represented by over 500 branches with more than 10,000 stops. About 332,000 passenger-agents are modelled to represent the demand for the TTC in the AM peak period. A learning-based departure time and path choice model was adopted using the concept of mental models for the modelling of the transit assignment problem. The choice model parameters were calibrated such that the entropy of the simulated route loads was optimized with reference to the observed route loads, and validated with individual choices. A Parallel Genetic Algorithm engine was used for the parameter calibration process. The modelled route loads, based on the calibrated parameters, greatly approximate the distribution underlying the observed loads. 75% of the exact sequence of transfer point choices were correctly predicted by the off-stop/on-stop choice mechanism. The model predictability of the exact sequence of route transfers was about 60%. In this application, transit passengers were assumed to plan their transit trip based on their experience with the transportation network; with no prior (or perfect) knowledge of service performance.     

Scheduling considerations for a branching transit route

This paper addresses the impacts of different scheduling alternatives for a branching transit route. It examines different schedule alternatives that might be used to optimize the route performance in terms of the passenger traveling time distributed among branch passengers and trunk‐line passengers. The schedule alternatives considered include transit vehicle allocation to different branches, offset shifting across vehicles on different branches, and vehicle holding (slack time) in the transit vehicle schedule. With these variables, several vehicle schedules are devised and examined based on a wide variety of possible passenger boarding scenarios using deterministic service models. Test outcomes provide general conclusions about the performance of the strategies. Vehicle assignment leading to even headways among branches is generally preferred for the case of low passenger demand. However, when passenger demand is high, or the differences between the passenger demands on branches are significant, unequal vehicle assignment will be helpful to improve the overall route performance. Holding, as a proactive strategy in scheduling, has the potential to be embedded into the schedule as a type of slack time, but needs further evidence and study to determine the full set of conditions where it may be beneficial. Offset shifting does not show sufficient evidence to be an efficient strategy to improve route performance in the case of low or high passenger demand.     

Does crowding affect the path choice of metro passengers?

This paper investigates crowding effect on the path choice of metro passengers. We show people reroute not only to avoid the delay from crowding but also to evade crowding itself. More specifically, a logit model fits best when it uses the transit delay from crowding as well as the passenger load of a connection in addition to the conventional explanatory variables. Also, we demonstrate that crowding decreases the overall welfare of metro passengers. The model is tested on the real path choice data acquired by the recent algorithm by Hong et al. (2015) known to detect the real path choice from Smart Card data in more than 90% of the cases.     

An activity-based approach for scheduling multimodal transit services

This paper proposes a new activity-based transit assignment model for investigating the scheduling (or timetabling) problem of transit services in multi-modal transit networks. The proposed model can be used to generate the short-term and long-term timetables of multimodal transit lines for transit operations and service planning purposes. The interaction between transit timetables and passenger activity-travel scheduling behaviors is captured by the proposed model, as the activity and travel choices of transit passengers are considered explicitly in terms of departure time choice, activity/trip chain choices, activity duration choice, transit line and mode choices. A heuristic solution algorithm which combines the Hooke–Jeeves method and an iterative supply–demand equilibrium approach is developed to solve the proposed model. Two numerical examples are presented to illustrate the differences between the activity-based approach and the traditional trip-based method, together with comparison on the effects of optimal timetables with even and uneven headways. It is shown that the passenger travel scheduling pattern derived from the activity-based approach is significantly different from that obtained by the trip-based method, and that a demand-sensitive (with uneven headway) timetable is more efficient than an even-headway timetable.     

A new schedule-based transit assignment model with travel strategies and supply uncertainties

This paper proposes a new scheduled-based transit assignment model. Unlike other schedule-based models in the literature, we consider supply uncertainties and assume that users adopt strategies to travel from their origins to their destinations. We present an analytical formulation to ensure that on-board passengers continuing to the next stop have priority and waiting passengers are loaded on a first-come-first-serve basis. We propose an analytical model that captures the stochastic nature of the transit schedules and in-vehicle travel times due to road conditions, incidents, or adverse weather. We adopt a mean variance approach that can consider the covariance of travel time between links in a space–time graph but still lead to a robust transit network loading procedure when optimal strategies are adopted. The proposed model is formulated as a user equilibrium problem and solved by an MSA-type algorithm. Numerical results are reported to show the effects of supply uncertainties on the travel strategies and departure times of passengers.     

Learning-based framework for transit assignment modeling under information provision

The modeling of service dynamics has been the focus of recent developments in the field of transit assignment modeling. The emerging focus on dynamic service modeling requires a corresponding shift in transit demand modeling to represent appropriately the dynamic behaviour of passengers and their responses to Intelligent Transportation Systems technologies. This paper presents the theoretical development of a departure time and transit path choice model based on the Markovian Decision Process. This model is the core of the MIcrosimulation Learning-based Approach to TRansit Assignment. Passengers, while traveling, move to different locations in the transit network at different points in time (e.g. at stop, on board), representing a stochastic process. This stochastic process is partly dependent on the transit service performance and partly controlled by the transit rider’s trip choices. This can be analyzed as a Markovian Decision Process, in which actions are rewarded and hence passengers’ optimal policies for maximizing the trip utility can be estimated. The proposed model is classified as a bounded rational model, with a constant utility term and a stochastic choice rule. The model is appropriate for modeling information provision since it distinguishes between individual’s experience with the service performance and information provided about system dynamics.     

Dynamic stochastic transit assignment with explicit seat allocation model

This paper proposes a stochastic dynamic transit assignment model with an explicit seat allocation process. The model is applicable to a general transit network. A seat allocation model is proposed to estimate the probability of a passenger waiting at a station or on-board to get a seat. The explicit seating model allows a better differentiation of in-vehicle discomfort experienced by sitting and standing passengers. The paper proposes simulation procedures for calculating the sitting probability of each type of passengers. A heuristic solution algorithm for finding an equilibrium solution of the proposed model is developed and tested. The numerical tests show significant influences of the seat allocation model on equilibrium departure time and route choices of passengers. The proposed model is also applied to evaluate the effects of an advanced public transport information system (APTIS) on travellers’ decision-making.     

An improved headway-based holding strategy for bus transit

Abstract

This paper presents an improved headway-based holding strategy integrating bus transit travel and dwelling time prediction. A support vector machine-based (SVM) model is developed to predict the baseline travel and dwell times of buses based on recent data. In order to reduce prediction errors, an adaptive algorithm is used together with real-time bus operational information and estimated baseline times from SVM models. The objective of the improved holding strategy is to minimize the total waiting times of passengers at the current stop and at successive stops. Considering the time-varying features of bus running, a ‘forgetting factor’ is introduced to weight the most recent data and reduce the disturbance from unexpected incidents. Finally, the improved holding strategy proposed in this study is illustrated using the microscopic simulation model Paramics and some conclusions are drawn.