Railway station choice modelling: a review of methods and evidence

Since the first railway station choice studies of the 1970s, a substantial body of research on the topic has been completed, primarily in North America, the U.K. and the Netherlands. With many countries seeing sustained growth in rail passenger numbers, which is forecast to continue, station choice models have an important role to play in assessing proposals for new stations or service changes. This paper reviews the modelling approaches adopted, the factors found to influence station choice and the application of models to real-world demand forecasting scenarios. A consensus has formed around using the closed-form multinomial logit and nested logit models, with limited use of more advanced simulation-based models, and the direction effects of a range of factors have been consistently reported. However, there are questions over the validity of applying non-spatial discrete choice models to a context where spatial correlation will be present, in particular with regard to the models’ ability to adequately represent the abstraction behaviours resulting from competition between stations. Furthermore, there has been limited progress towards developing a methodology to integrate a station choice element into the aggregate models typically used to forecast passenger demand for new stations.     

Air itinerary shares estimation using multinomial logit models

ABSTRACT

The main goal of this study is the development of an aggregate air itinerary market share model. In order to achieve this, multinomial logit models are applied to distribute the city-pair passenger demand across the available itineraries. The models are developed at an aggregate level using open-source booking data for a large group of city-pairs within the US air transport system. Although there is a growing trend in the use of discrete choice models in the aviation industry, existing air itinerary share models are mostly focused on supporting carrier decision-making. Consequently, those studies define itineraries at a more disaggregate level using variables describing airlines and time preferences. In this study, we define itineraries at a more aggregate level, i.e. as a combination of flight segments between an origin and destination, without further insight into service preferences. Although results show some potential for this approach, there are challenges associated with prediction performance and computational intensity.     

Model development for metro station access mode choice

This paper analyses the behaviour of metro users in choosing their access mode to a metro station. Multinominal logit models with satisfactory predictive power were developed for access mode choice on the basis of data collected by interviewing metro users at existing metro stations. A population segmentation approach was adopted and models referring to individuals having the same set of alternative access modes were developed. Trip purpose was found to have significant effects on the access mode choice. Thus, for each population segment different models are proposed for work and education and other trip purpose. Various conclusions concerning the importance of the variables included in the proposed models were drawn through comparisons carried out across the models.     

A comparative study of aggregate and disaggregate gravity models using Seoul metropolitan subway trip data

Abstract

Existing origin constrained and doubly constrained gravity models have not been compared, theoretically or empirically, in terms of their forecasting power. Due to the newly advanced technology of intelligent transport systems, the expanded data presently available have made various models more comparable in terms of forecasting power. This paper uses archived automatic passenger counting (APC) data for urban rail in the Seoul metropolitan area. The APC data contains information about each trip's origin, destination, ticket type, fare, and distance on a daily basis. The objective of this paper is to compare the goodness-of-fit of aggregate and disaggregate gravity modeling using these data. A Hyman aggregate gravity model is used as the aggregate model without the spatial effect. The disaggregate model adopts a multinomial logit as the destination choice model with the spatial effect. In general, while the formulation of aggregate and disaggregate gravity model models are similar, the calibration and parameter estimation methods of the two models are different. As a result, this empirical study demonstrates that the variation in goodness-of-fit and forecasting power largely depends on the estimation method and selected variables. The forecasting power of the disaggregate modeling approach outperforms that of the aggregate model. This paper further confirms that spatial arrangement plays important roles in gravity modeling.     

The federal republic of Germany: the federal ministry of transport

The continued liberalization of the airline market in the European Community offers the prospect of competition in terms of price, frequency of service and many other attributes. Understanding these factors that influence passenger demand and incorporating them into a demand function poses a substantial theoretical and empirical challenge. A basis for estimating demand functions is the logit model. This paper examines the value of a logit share model in explaining how the share of airlines between airports varies as a function of measured and unmeasured attributes. It shows how demand functions are necessary for understanding a competitive market and evaluates the progress to date in identifying accurate coefficients of the logit function.     

Further evidence on the temporal stability of disaggregate travel demand models

Current evidence on the transferability of disaggregate travel demand models is inconclusive. Adding to this body of research, the present analysis focuses upon the temporal characteristics of work trip behavior in the San Francisco Bay Area. Using before and after data sets associated with the BART Impact Travel Study, multinomial logit models of work trip modal choice are estimated. The results indicate that the general form and the coefficient estimates of a pre BART model are transferable in time. Moreover, when updated to reflect BART's presence, the model's predictive success and its implied elasticity measures are generally accurate, relative to those implied by reestimating the entire model on post BART data. Finally, as economic theory would predict, elasticity measures of the service related variables were found to increase over time.     

Development of destination choice models for pedestrian travel

Most research on walking behavior has focused on mode choice or walk trip frequency. In contrast, this study is one of the first to analyze and model the destination choice behaviors of pedestrians within an entire region. Using about 4500 walk trips from a 2011 household travel survey in the Portland, Oregon, region, we estimated multinomial logit pedestrian destination choice models for six trip purposes. Independent variables included terms for impedance (walk trip distance), size (employment by type, households), supportive pedestrian environments (parks, a pedestrian index of the environment variable called PIE), barriers to walking (terrain, industrial-type employment), and traveler characteristics. Unique to this study was the use of small-scale destination zone alternatives. Distance was a significant deterrent to pedestrian destination choice, and people in carless or childless households were less sensitive to distance for some purposes. Employment (especially retail) was a strong attractor: doubling the number of jobs nearly doubled the odds of choosing a destination for home-based shopping walk trips. More attractive pedestrian environments were also positively associated with pedestrian destination choice after controlling for other factors. These results shed light on determinants of pedestrian destination choice behaviors, and sensitivities in the models highlight potential policy-levers to increase walking activity. In addition, the destination choice models can be applied in practice within existing regional travel demand models or as pedestrian planning tools to evaluate land use and transportation policy and investment scenarios.     

Solving the station-based one-way carsharing network planning problem with relocations and non-linear demand

One-way station-based carsharing systems allow users to return a rented car to any designated station, which could be different from the origin station. Existing research has been mainly focused on the vehicle relocation problem to deal with the travel demand fluctuation over time and demand imbalance in space. However, the strategic planning of the stations’ location and their capacity for one-way carsharing systems has not been well studied yet, especially when considering vehicle relocations simultaneously. This paper presents a Mixed-integer Non-linear Programming (MINLP) model to solve the carsharing station location and capacity problem with vehicle relocations. This entails considering several important components which are for the first time integrated in the same model. Firstly, relocation operations and corresponding relocation costs are taken into consideration to address the imbalance between trip requests and vehicle availability. Secondly, the flexible travel demand at various time steps is taken as the input to the model avoiding deterministic requests. Thirdly, a logit model is constructed to represent the non-linear demand rate by using the ratio of carsharing utility and private car utility. To solve the MINLP model, a customized gradient algorithm is proposed. The application to the SIP network in Suzhou, China, demonstrates that the algorithm can solve a real world large scale problem in reasonable time. The results identify the pricing and parking space rental costs as the key factors influencing the profitability of carsharing operators. Also, the carsharing station location and fleet size impact the vehicle relocation and carsharing patronage.     

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.     

Trip chaining as a barrier to the propensity to use public transport

Trip chaining is a growing phenomenon in travel and activity behaviour. Individuals increasingly seek out opportunities to minimise the amount of travel required as part of activity fulfilment, given the competing demands on time budgets and their valuation of travel time savings. This search for ways of fulfilling (more) activities with less travel input has produced a number of responses, one of which is trip chaining. A particularly important policy implication of trip chaining is the potential barrier it creates in attracting car users to switch to public transport. This paper seeks to improve our understanding of trip chaining as a barrier to public transport use. A series of discrete choice models are estimated to identify the role that socio-economic and demographic characteristics of households have on the propensity to undertake trip chains of varying degrees of simplicity/complexity that involve use of the car or public transport with an embedded commuting or non-commuting primary purpose. Multinomial logit, nested logit and random parameter logit models are developed and contrasted to establish the gains in relaxing the strict conditions of the multinomial logit model.     

Impacts of highly automated vehicles on travel demand: macroscopic modeling methods and some results

Automated vehicles (AV) will change transport supply and influence travel demand. To evaluate those changes, existing travel demand models need to be extended. This paper presents ways of integrating characteristics of AV into traditional macroscopic travel demand models based on the four-step algorithm. It discusses two model extensions. The first extension allows incorporating impacts of AV on traffic flow performance by assigning specific passenger car unit factors that depend on roadway type and the capabilities of the vehicles. The second extension enables travel demand models to calculate demand changes caused by a different perception of travel time as the active driving time is reduced. The presented methods are applied to a use case of a regional macroscopic travel demand model. The basic assumption is that AV are considered highly but not fully automated and still require a driver for parts of the trip. Model results indicate that first-generation AV, probably being rather cautious, may decrease traffic performance. Further developed AV will improve performance on some parts of the network. Together with a reduction in active driving time, cars will become even more attractive, resulting in a modal shift towards car. Both circumstances lead to an increase in time spent and distance traveled.

     

Evaluating light rail sketch planning: actual versus predicted station boardings in Phoenix

In recent years, transit planners are increasingly turning to simpler, faster, and more spatially detailed “sketch planning” or “direct demand” models for forecasting rail transit boardings. Planners use these models for preliminary review of corridors and analysis of station-area effects, instead of or prior to four-step regional travel demand models. This paper uses a sketch-planning model based on a multiple regression originally fitted to light-rail ridership data for 268 stations in nine U.S. cities, and applies it predictively to the Phoenix, Arizona light-rail starter line that opened in December, 2008. The independent variables in the regression model include station-specific trip generation and intermodal–access variables as well as system-wide variables measuring network structure, climate, and metropolitan-area factors. Here we compare the predictions we made before and after construction began to pre-construction Valley Metro Rail predictions and to the actual boardings data for the system’s first 6 months of operations. Depending on the assumed number of bus lines at each station, the predicted total weekday ridership ranged from 24,767 to 37,907 compared with the average of 33,698 for the first 6 months, while the correlation of predicted and observed station boardings ranged from r = 0.33 to 0.47. Sports venues, universities, end-of-line stations, and the number of bus lines serving each station appear to account for the major over- and under-predictions at the station level.     

Estimating the long-term effects of different passenger car technologies on energy/fuel consumption and emissions of greenhouse gases in Europe

Abstract

A stated preference (SP) experiment of car ownership was conducted in Mumbai Metropolitan Region (MMR) of Maharashtra in India. A full factorial experiment was designed to considering various attributes such as travel time, travel cost, projected household income, car loan payment and servicing cost. Data on 357 individuals were collected which resulted in 3213 observations for the calibration of the work trip and recreational trip car ownership models. The car ownership alternatives considered 0, 1 and 2 cars. A multinomial logit framework was used to develop the car ownership model taking the household as a decision unit. The specification and results of the SP car ownership model are discussed. The observed and predicted values matched reasonably when the validity of the SP car ownership model was tested against revealed preference (RP) data. The car ownership models developed in this study exhibit a satisfactory goodness of fit. It is concluded that the SP modelling approach can be successfully used for modelling car ownership decisions of households in developing countries.     

Forecasting Travel Demand with Alternatively Structured Models of Trip Frequency

Abstract

This paper develops alternatively structured trip frequency/generation models, and investigates their forecast performance. The first model presented is the simple linear model with a discussion of its theoretical shortcomings. Models that address, in a progressive fashion, the underlying shortcomings of the linear model are then presented. These models are namely the truncated normal model, the Poisson model, the negative binomial model, and an ordered logit model. The modeling unit employed in the study is the individual. The models are assessed by how closely they are able to replicate trips produced by each individual in the dataset, and by each traffic zone. This assessment of performance in prediction is conducted on an estimation dataset collected in the Toronto Region in 1986, and on an independent dataset collected in the same geographic region, 10 years later, in 1996. The results show that, notwithstanding the simplicity of the simple linear model and its lack of an explicit underlying travel behavioral theory, it predicts travel in the base and forecast years with less error compared to any of the more complex models.     

Real time transit demand prediction capturing station interactions and impact of special events

Demand for public transportation is highly affected by passengers’ experience and the level of service provided. Thus, it is vital for transit agencies to deploy adaptive strategies to respond to changes in demand or supply in a timely manner, and prevent unwanted deterioration in service quality. In this paper, a real time prediction methodology, based on univariate and multivariate state-space models, is developed to predict the short-term passenger arrivals at transit stations. A univariate state-space model is developed at the station level. Through a hierarchical clustering algorithm with correlation distance, stations with similar demand patterns are identified. A dynamic factor model is proposed for each cluster, capturing station interdependencies through a set of common factors. Both approaches can model the effect of exogenous events (such as football games). Ensemble predictions are then obtained by combining the outputs from the two models, based on their respective accuracy. We evaluate these models using data from the 32 stations on the Central line of the London Underground (LU), operated by Transport for London (TfL). The results indicate that the proposed methodology performs well in predicting short-term station arrivals for the set of test days. For most stations, ensemble prediction has the lowest mean error, as well as the smallest range of error, and exhibits more robust performance across the test days.     

A tour-based model of travel mode choice

This paper presents a new tour-based mode choice model. The model is agent-based: both households and individuals are modelled within an object-oriented, microsimulation framework. The model is household-based in that inter-personal household constraints on vehicle usage are modelled, and the auto passenger mode is modelled as a joint decision between the driver and the passenger(s) to ride-share. Decisions are modelled using a random utility framework. Utility signals are used to communicate preferences among the agents and to make trade-offs among competing demands. Each person is assumed to choose the best combination of modes available to execute each tour, subject to auto availability constraints that are determined at the household level. The households allocations of resources (i.e., cars to drivers and drivers to ride-sharing passengers) are based on maximizing overall household utility, subject to current household resource levels. The model is activity-based: it is designed for integration within a household-based activity scheduling microsimulator. The model is both chain-based and trip-based. It is trip-based in that the ultimate output of the model is a chosen, feasible travel mode for each trip in the simulation. These trip modes are, however, determined through a chain-based analysis. A key organizing principle in the model is that if a car is to be used on a tour, it must be used for the entire chain, since the car must be returned home at the end of the tour. No such constraint, however, exists with respect to other modes such as walk and transit. The paper presents the full conceptual model and estimation results for an initial empirical prototype. Because of the complex nature of the model decision structure, choice probabilities are simulated from direct generation of random utilities rather than through an analytical probability expression.     

A comparative study of a hybrid Logit–Fratar and neural network models for trip distribution: case of the city of Isfahan

This paper introduces a new procedure to forecast the future O/D demand. It is a hybrid of logit and Fratar model. The hybrid model has the long run, policy sensitive, characteristic of a logit model, calibrated at sector‐level with little/no zero O/D cells. This feature, joint with a Fratar‐type operation at zonal level within a sector, gives a better performance to this model than either of the two types of the models alone. The performance of the hybrid model is contrasted with a neural network model, and shows encouraging results in a real case. Copyright © 2010 John Wiley & Sons, Ltd.     

European passenger rail reforms: A comparative assessment of the emerging models

This paper provides a comprehensive framework for analysing and comparing the various institutional models and regulatory arrangements that have recently emerged from the restructuring of European passenger rail industries. The framework identifies seven key dimensions that can be used to describe these models and arrangements. It shows how these dimensions are connected with one another in a complex web of interactions. Empirical material gathered on the passenger rail industries of five EU Member States, i.e. Belgium, France, Germany, the UK, and Sweden is used, as well as extrapolation to highlight the costs and benefits (incentive properties) associated with alternative arrangements and contract features and to show how certain dimensions can be traded-off against one another to optimize the industry's performance.     

Updating predictive accident models of modern rural single carriageway A-roads

Abstract

Reliable predictive accident models (PAMs) are essential to design and maintain safe road networks, and yet the models most commonly used in the UK were derived using data collected 20 to 30 years ago. Given that the national personal injury accident total fell by some 30% in the last 25 years, while road traffic increased by over 60%, significant errors in scheme appraisal and evaluation based on the models currently in use seem inevitable. In this paper, the temporal transferability of PAMs for modern rural single carriageway A-roads is investigated, and their predictive performance is evaluated against a recent data set. Despite the age of these models, the PAMs for predicting the total accidents provide a remarkably good fit to recent data and these are more accurate than models where accidents are disaggregated by type. The performance of the models can be improved by calibrating them against recent data.     

Estimating GEV models with censored data

We examine the problem of estimating parameters for Generalized Extreme Value (GEV) models when one or more alternatives are censored in the sample data, i.e., all decision makers who choose these censored alternatives are excluded from the sample; however, information about the censored alternatives is still available. This problem is common in marketing and revenue management applications, and is essentially an extreme form of choice-based sampling. We review estimators typically used with GEV models, describe why many of these estimators cannot be used for these censored samples, and present two approaches that can be used to estimate parameters associated with censored alternatives. We detail necessary conditions for the identification of parameters associated exclusively with the utility of censored alternatives. These conditions are derived for single-level nested logit, multi-level nested logit and cross-nested logit models. One of the more surprising results shows that alternative specific constants for multiple censored alternatives that belong to the same nest can still be separately identified in nested logit models. Empirical examples based on simulated datasets demonstrate the large-sample consistency of estimators and provide insights into data requirements needed to estimate these models for finite samples.