This study investigates the spatiotemporal relationship between land-use patterns and transit ridership along Bangkok’s MRT Blue Line using stepwise and geographically weighted regression (GWR) analysis. Drawing on ridership data and analyzing land-use composition within 500-meter buffers, the study identifies residential, commercial-office, and transport utility land uses as the primary predictors of both weekday and weekend ridership. Additionally, malls and markets emerge as significant predictors during late evening weekday and afternoon weekend periods, reflecting non-commuting travel behavior. The GWR results highlight substantial spatial heterogeneity, demonstrating how the influence of land use on ridership varies across time and location.
Railway systems are complex and consist of many subsystems working together to ensure smooth and timely operation. Delays that cause deviations from the scheduled operations can lead to cascading issues throughout the railway network. These delays can be a result of a large number of causes, from adverse weather to random equipment failure. Delay prediction is crucial in the mitigation of delay effects. This paper proposes a data-driven approach to the prediction of railway delays. Applying Recursive Neural Networks in this data-driven approach allows it to exploit the sequential nature of train operations without having to make intermediate predictions common in event-driven approaches. Raw scheduling data are processed to compute features relevant to the analysis of delay and its propagation. In the proposed method, each row of data is weighted according to its temporal distance to the prediction horizon, assigning increased importance to more recent events. These time-weighted data are analysed at different scales via the component long-short-term memory (LSTM) models in the proposed ensemble model. Data of a 1-year time period from the Historical Service Records of the British Railway were used to train and test the proposed time-weighted ensemble LSTM architecture. The proposed architecture showed an improved performance compared to other data-driven models implemented as benchmarks, outperforming them by achieving a root mean squared error of 0.27 min, a mean absolute error of 0.17 min as well as a coefficient of determination (R-2) value of 0.9875.
This article examines whether passenger fares in Greek coastal shipping remain distance-driven after market liberalisation. Using a harmonised 2025 cross-section of 52 non-PSO routes (i.e. routes not operated under Public Service Obligation—PSO—contracts), we measure the association between the lowest economy-class passenger fare and distance, compare 11 curve families, and derive a route-level fare benchmark. A no-intercept power function emerges as the empirically preferred specification. For high-speed services (S), distance explains 99.3
In the off-peak hours, rolling stock is put on shunting yards. Since the number of used trains increases and the size of the shunting yards remains the same, it becomes harder and harder to find a feasible schedule for the shunting yards. The Dutch Railways (NS) is developing a software package that is capable of finding feasible schedules, but these often contain a high number of shunt movements, which is undesirable. In this paper, we propose a method that can reduce the number of shunt movements in existing feasible schedules by repeatedly rescheduling trains. We repeatedly reschedule one train at a time using the observation that finding a new schedule is equivalent to finding where and when a train will park. We build a model that allows us to find feasible parking locations and time intervals while taking into account the length of the tracks. By carefully looking at the position of the trains on the track, we are also able to avoid a special type of movement, called move-up movements. We apply a variant of a shortest path with time windows algorithm on this model, resulting in a schedule with the smallest number of movements for the rescheduled train. We then extend this approach to reschedule two trains simultaneously. We conclude with experiments on real-world data and a description of other possible use cases for the developed algorithm.
In this paper, we consider the electric vehicle scheduling problem (e-VSP): a set of trips corresponding to a given timetable have to be driven by a set of electric buses with limited capacity. This problem, like many other planning problems, boils down to assigning to each bus a subset of the trips with the obvious side constraint that the selected subset can be feasibly driven by this single bus; such a feasible subset is called a vehicle schedule. If we know all possible vehicle schedules, we can select the best set of these by solving an integer linear program (ILP). This idea has inspired many researchers to the heuristic of finding a decent subset of all vehicle schedules using the technique of column generation and then solve the ILP. For the e-VSP, this approach leads to good solutions, but there is still room for improvement. Instead of using column generation, we apply simulated annealing to find the subset of vehicle schedules that we use as input for the ILP. For the e-VSP, this leads to better solutions. Moreover, this approach, which we call simulated annealing with recombination through ILP, is generally applicable and has as a clear advantage that we do not have to solve the pricing problem, because this approach increases the application possibilities and takes far less time.
This paper presents a scalable methodology for real-time transfer synchronization in urban bus networks, using online stochastic optimization (OSO). The approach integrates three key components. First, an offline arc-flow model captures all control tactics—hold, speedup, and skip-stop—for a main line and its feeder connections, using a graph-based representation over a fixed control horizon. Second, the Regret (R) algorithm operates in real time within an OSO framework, leveraging the offline model to evaluate multiple stochastic scenarios and to select robust control tactics. Third, a network-wide simulator (NWS) integrates the full OSO framework and re-optimizes decisions dynamically at each bus departure from any stop, allowing the coordination of multiple interconnected lines. The NWS is applied to the transit network of Laval, Canada, using historical vehicle positions and smart card validations to replicate real-time stochastic conditions. Results show significant improvements in both passenger travel and transfer times across a variety of network structures, highlighting the scalability and applicability of real-time transfer synchronization for urban multi-line transit networks.
The aim of this paper is to develop an evaluation index for resident accessibility to public transport (PTAIR). This index expands over previous measures of public transport accessibility by incorporating additional parameters, specifically spatial resident distribution and transport capacity, with the objective of integrating service quality into accessibility assessment. Research that simultaneously accounts for population distribution and transport capacity as key indicators of accessibility remains limited. To address this gap, the proposed index was applied to the New City of Sidi-Abdallah (NCSA) in Algiers as a case study. Accessibility levels were evaluated and represented through maps generated using a geographic information system. The results reveal a significant mismatch between the transport system and the spatial distribution of residents, particularly in newly developed collective residential neighborhoods where access to public transport is either poor or very poor. These findings suggest that the PTAIR index can serve as a valuable tool for improving public transport planning by aligning accessibility more closely with resident distribution.
This paper revisits approaches and methods of Transit Performance Evaluation (TPE) in the context of developing countries, where existing TPE measures are found to have certain deficiencies. TPE approaches and measures are discussed related to efficiency and effectiveness, service quality, Level of Service (LOS), community perspectives and a combination of them. Efficiency and effectiveness measures are related to administration and logistics or system functions and operations. Service quality considers passenger perception towards offered services. The LOS approach allows a comparison based on indicator thresholds subject to transit systems having equitable environments. It is observed that community perspectives are given lower importance. This review helps to identify areas which need to be improved in the existing frameworks. The proposed framework is an index-based framework and is comprehensive, inclusive, has universal applicability and considers the complete ensemble of transit systems. It includes stakeholders’ perspectives to develop three indices covering operational, service quality, and community and society domains. While developing the framework, data availability, variability and heterogeneity are taken into consideration.
This study proposes a comprehensive and systematic comparison between Demand-Responsive Transport (DRT) and Public Transport (PT) in a real-world context, Canberra, Australia, by replacing a local bus line with DRT. The conducted comparison considers key indicators such as the number of vehicles, travel distance, operational costs, fuel consumption, and passenger travel time. To enhance the decision-making process, a multi-objective approach is used to simultaneously optimise operational costs, environmental impacts, and passenger inconvenience. Various weight combinations are used to explore the trade-offs amongst these objectives. The proposed model is simulated on a real-world case study in Canberra, using the public transport smart card data from 2019 (January) to 2021 (July). To analyse the impact of daily variations in demand, three operation shifts are simulated. The findings reveal that DRT is more flexible, efficient, and environmentally sustainable in low-demand scenarios, while PT excels in cost-efficiency as demand increases. It is observed that changing the weights associated with the objectives can influence operational costs by 25
In many developing cities, urban-rail transit systems are planned under long-term master plans but often lack detailed evaluations of how network expansion changes accessibility by public transport. This study assesses the impact of railway network expansion using a cumulative accessibility index derived from GTFS-based travel times. Three scenarios are compared: the current network, the original M-MAP plan, and the proposed M-MAP2 Blueprint. The results show that the M-MAP2 Blueprint increases citywide accessibility by approximately 10
Diligent railway track irregularities maintenance scheduling is important for a safe rail traffic operation. Machine learning classifiers based on vehicle response have been shown to be an effective approach for track irregularities assessment. However, when simultaneously assessing several types of track irregularities within a given track section, loss of interpretation becomes a prevalent issue as the track irregularities assessment and the accompanying machine learning classification become more complicated. The present work examines the use of machine learning classification for combined track irregularities assessment and the subsequent result interpretation using Shapley Additive Explanation (SHAP) and Accumulated Local Effects (ALE). Testing results of the trained classification models show a high accuracy value, i.e., higher than 92
Equity in public transit accessibility is increasingly recognized as a critical planning objective, yet existing evaluation tools often fall short in interpretability and operational relevance. This paper introduces a practical and reproducible toolbox designed to assess vertical equity in spatial accessibility, focusing on disparities between potentially vulnerable and non-vulnerable populations. Grounded in a Rawlsian equity perspective, the toolbox combines two accessibility indicators, public transit accessibility and urban opportunity accessibility, with three intuitive equity metrics and two graphical diagnostic tools. To demonstrate the applicability of the toolbox, we use the indicators to assess the current equity situation of spatial accessibility on the territory of the Montreal regional transport authority. The preliminary investigation reveals the existence of a specific study area experiencing significant levels of inequity in 2023. In this area, vulnerable populations experience 12–19
Ridesharing allows owners of private cars to share these with other travelers, filling up vacant seats. Based on a real case from Norway, we define and study the Ridesharing Problem with Flexible Riders and Fixed Lines (RPFRFL), where flexible riders can act either as passengers or drivers that can pick up other passengers. Moreover, the RPFRFL also considers fixed (bus) lines, to which drivers can deliver picked up passengers at a terminal for public transportation to their final destinations. We propose a mixed integer programming model and a path flow solution method to solve this problem. Based on a large number of test instances for our case study, we study the effects of considering flexible riders and the integration with fixed bus lines and show, for example, that more than 10
The emphasis on the efficient utilization of public transportation resources has become particularly relevant in recent years due to the post-pandemic fluctuation in public transportation usage and the rise in operational costs. The analysis of transportation usage rates provides valuable insights into the efficiency of the service, offering an indicator that integrates actual demand with the capacity. This study aims to develop a methodology for analyzing the occupancy rate from large-scale datasets to identify gaps between supply and demand in public transportation. Leveraging the spatio-temporal granularity of data from Automatic People Counting (APC) systems and relying on the Generalized Linear Mixed Effects Model and the Generalized Mixed-Effect Random Forest, in this study we propose a methodology for analyzing factors determining low occupancy rates. The model’s results are examined at both the segment and ride levels. Initially, the analysis focuses on identifying segments more likely associated with low occupancy rates, understanding factors influencing the probability of having low occupancy rates, and exploring their relationships. Subsequently, the analysis extends to the temporal distribution of low-occupancy-rate situations, encompassing its impact on the entire journey. The proposed methodology is applied to analyze APC data, provided by the company responsible for public transport management in Milan, on a radial route of the surface transportation network.
The overcrowding scenarios in suburban railway stations in India during peak hours have necessitated the need for an intervention by the Indian Railways. To address this problem, the Indian Railways put forward a change in the design of the platforms by introducing double-discharge platforms to handle arriving and departing commuters. The intended impact of the revised design was to reduce overcrowding in platforms and footbridges and mitigate potential accidents which can occur as a result of overcrowding. We perform a multi-method analysis of this policy intervention to explain the implications of the double-discharge platforms on crowd management in suburban railway stations. In this regard, a game-theoretic analysis of the double-discharge platform theoretically illustrates how overcrowding is managed with the intervention. Further, a pedestrian-simulation analysis is also performed to understand how the crowding scenarios evolve in suburban railway stations with the intervention. The analysis brings about the effectiveness of the double-discharge platform solution using multiple lenses.
Bus transportation offers an effective alternative to private vehicles, promoting a more sustainable service and playing a crucial role in urban mobility. However, different factors can affect bus reliability in service operations. The difference between actual and scheduled arrival times at each stop is used to assess reliability. It accounts for both bus delays, when the vehicle runs behind schedule, and early arrivals, when the vehicle runs ahead of schedule. This paper proposes an approach to assess bus service reliability based on Global Navigation Satellite System (GNSS) data from connected vehicles. The initial step includes estimating bus arrival time at each bus stop. The second step consists of comparing the actual bus arrival time with the expected scheduled time. The study area is located at the main campus of the University of Campinas in Brazil. GNSS data were obtained from a monitoring system called Main Module IoT (MMIoT), which is responsible for providing bus trajectory information. In addition, occupancy data from a bus passenger counting (BPC) system installed in the same buses are used to evaluate the impact of bus occupancy on travel time reliability. Results from input data analysis revealed significant bus delays, especially during peak hours. On-time performance (OTP) is used as an indicator of bus service reliability. Moreover, Pearson correlation coefficients are used to examine the association between service reliability and bus occupancy on campus. Finally, the results provide insights into improving bus service reliability and efficiency on campus, thereby potentially increasing passenger use and satisfaction.
This paper introduces a novel Logistic Burr mixture autoregressive model (LBMAR) for two key bus scheduling and operation problems. First, the LBMAR model effectively analyzes and quantifies the individual and joint impacts of contributing factors on the time-varying bimodal distribution of bus section travel time. Unlike traditional approaches that examine factors between bus sections, our study investigates their impact on the validated bimodal variations in travel time for various operating environments of bus sections. Second, the LBMAR model jointly accounts for contributing factors in point and interval predictions of bus section travel time. Validated using six months of data and 169 bus section travel time observations with passenger information, we meticulously analyze four key contributing factors: day of the week, time of the day, bus occupancy changes, and neighboring bus section's travel time variability. The results show a remarkable correlation between bus occupancy changes and bimodal variations in bus section travel time. Moreover, the LBMAR model exhibits significant improvements in point and interval predictions, particularly for high to medium levels of variations in bus section travel time. These findings have profound implications for real-time bus operation management. By effectively identifying, quantifying, and managing the impact of these contributing factors, bus transportation operators can make informed decisions to optimize their operations, resulting in more efficient and reliable services.
The rapid growth of e-commerce and urbanization intensity causes last-mile delivery challenges, including congestion, high emissions, and inefficiencies. Crowdshipping—leveraging public transport commuters to deliver parcels—emerges as a promising solution that aligns with urban sustainability goals. This study investigates the behavioral drivers of crowdshipping participation among train commuters in Australia using a mixed-methods approach grounded in the theory of planned behavior. A quantitative survey of 368 respondents is analyzed through binary logistic regression, revealing that factors such as environmental concern, income reliability, and social influence strongly predict willingness to participate. These findings are further enriched with semi-structured interviews of 19 logistics and mobility experts from Europe, Australia, North and South America, and Southeast Asia, which support transferability of the core design levers. Overall, the results underscore that integrating crowdshipping into public transport networks requires a focus on sustainable benefits, fair and clear compensation structures, and very little disruption to daily commutes. The study offers actionable recommendations for policymakers, public transport operators, and logistics providers aiming to develop integrated urban logistics solutions that effectively reduce congestion and environmental impact.
The current growth of the urban population in cities is leading to mobility challenges. The increase of private cars in the city centers has become unsustainable and changes in the efficiency and diversity of public transportation are needed. Urban buses are regarded as a viable and cost-effective solution, namely the bus rapid transit systems. Several methods have been studied and developed to design efficient urban bus networks. This work presents a mathematical programming model to design a bus network. The objective is to minimize the time that an average user spends inside the network, by defining a set of routes, with associated frequencies that satisfy the demand predicted for each origin–destination pair. It is then applied to a case study in the Barcelona Metropolitan Area, where the impact of the implementation of a bus rapid transit route in the existing network is studied. The results serve as a decision support to improve the network, by returning the set of routes and associated frequencies that minimize the target objective function. A decrease of 8.75 min in the time spent by an average user inside the network is achieved, representing a reduction of 15
Distributions of boarding and alighting passengers on metro platforms have a significant impact on train dwell times and, consequently, a metro system's capacity. The extent of passengers’ awareness of the platform layout at alighting stations can lead to different behaviour on the platform, which determines the boarder and alighter distribution. This research develops a model for boarder and alighter distributions on metro platforms, considering the heterogeneity in passengers’ awareness and the corresponding behaviour. The train passenger loading data of the Hammersmith City Line of London Underground are used for the model calibration and validation. The average absolute error of the calibration results at the car-level is only 1.2 passengers, and the maximum absolute error at the door-level is less than two passengers. It is estimated that, on average, 60
