Generation of synthetic datasets for discrete choice analysis

Despite the widespread use of synthetic data in discrete choice analysis, little is known about how the methodology used to generate synthetic datasets influences the properties of parameter estimates and the validity of results based on these estimates. That is, there are two potential sources of biases when using synthetic discrete choice data: (1) bias due to the method used to generate the dataset; and, (2) bias due to parameter estimation. The primary objective of this study is to examine bias due to the underlying data generation method. This study compares three methods for generating synthetic datasets and uses design of experiments and analysis of variance methods to investigate the ability to recover estimates for “true” logsum parameters for nested logit models. The method that uses nested logit probabilities to generate the chosen alternative results in unbiased parameter estimates. The method that is based on Gumbel error component approximations reveals that while the error components themselves are unbiased, subtle empirical identification problems can arise when these error components are combined with synthetically generated utility functions. The method that is based on normal error component approximations reveals that all logsum coefficients are biased upwards; the bias dramatically increases for those nests that have a low choice frequency and is most pronounced for those nests with high correlations among alternatives. Based on the results of the analysis, several recommendations for the generation of synthetic datasets for discrete choice analyses are provided.     

The accuracy of the multinomial logit model as an approximation to the multinomial probit model of travel demand

The multinomial probit model of travel demand is considerably more general but much less tractable than the better-known multinomial logit model. In an effort to determine the effects of using the relatively simple logit model in situations where the assumptions of probit modeling are satisfied but those of logit modeling are not, the accuracy of the multinomial logit model as an approximation to a variety of three-alternative probit models has been evaluated. Multinomial logit can give highly erroneous estimates of the choice probabilities of multinomial probit models. However, logit models appear to give asymptotically accurate estimates of the ratios of the coefficients of the systematic components of probit utility functions, even when the logit choice probabilities differ greatly from the probit ones. Large estimation data sets are not necessarily needed to enable likelihood ratio tests to distinguish three-alternative probit models from logit models that give seriously erroneous estimates of the probit choice probabilities. Inclusion of alternative-specific dummy variables in logit utility functions cannot be relied upon to reduce significantly the errors of logit approximations to the choice probabilities of probit models whose utility functions do not contain the dummies.     

A utility maximizing model of the demand for multi-destination non-work travel

The multinomial probit model of travel demand is considerably more general but much less tractable than the better-known multinomial logit model. In an effort to determine the effects of using the relatively simple logit model in situations where the assumptions of probit modeling are satisfied but those of logit modeling are not, the accuracy of the multinomial logit model as an approximation to a variety of three-alternative probit models has been evaluated. Multinomial logit can give highly erroneous estimates of the choice probabilities of multinomial probit models. However, logit models appear to give asymptotically accurate estimates of the ratios of the coefficients of the systematic components of probit utility functions, even when the logit choice probabilities differ greatly from the probit ones. Large estimation data sets are not necessarily needed to enable likelihood ratio tests to distinguish three-alternative probit models from logit models that give seriously erroneous estimates of the probit choice probabilities. Inclusion of alternative-specific dummy variables in logit utility functions cannot be relied upon to reduce significantly the errors of logit approximations to the choice probabilities of probit models whose utility functions do not contain the dummies.     

A decision support system for airport strategic planning

This paper describes an integrated set of models for the estimation of the capacity of an airfield and the associated delays. The aim is to develop a decision support tool suitable for airport planning at the strategic level. Thus, the emphasis is on obtaining reliable approximations to the quantities of interest quickly and with a limited set of inputs. The models account for the dynamic characteristics of airfield capacity and demand, as well as for some stochastic aspects of airfield operations. They are sensitive to airfield geometry, the operational characteristics of the airfield and of the local air traffic control system, and the characteristics of the local air traffic demand for airport access and services. Through its integrated structure, the decision support tool can account for interactions among operations at different parts of the airfield.     

A comparative analysis of US toll policy

In the US, there is a long tradition of toll roads, beginning with the Lancaster Turnpike that was built at the end of the 18th century connecting Philadelphia and Lancaster. There are currently more than 300 toll facilities in the US, which is probably the largest number of toll facilities in the world. These facilities represent a wide range of conditions, from hypercongested facilities in large metropolitan areas such as New York City to toll highways in rural areas. The toll structures are equally diverse, ranging from multi-tier price structures with frequent user, carpool, and time of day discounts; to simpler structures in which the only differentiation is made on the basis of the number of axles per vehicle. The toll rates are typically set by the agencies that operate or own the toll facilities. The rules or formulas by which these tolls are determined are not generally available to the public, though it is safe to say that toll decisions are made taking into account technical considerations, as well as the all important criterion of political acceptability. However, data on toll rates and how they change by vehicle types and by some other attributes are readily available.The overall objective of this paper is to analyze the toll data from various facilities across the US to gain insight into the overall factors affecting the tolls. A more specific objective is to assess—though in a rather approximate fashion—if the tolls by vehicle type, relative to each other, are appropriate and consistent with economic theory. This is achieved by comparing tolls to approximate indicators of road space consumption and pavement deterioration. The literature review confirmed that this is the first time such research has been conducted which is an important first step toward an analysis of the efficiency of current toll policies.The analyses in this paper are based on a random sample of all toll facilities across the US. The toll dataset, which include toll rates for passenger cars, busses, and three different truck types, is assembled mainly from the available information on the web sites of various toll agencies. After cleaning the data, the authors used econometric modeling to estimate a set of ordinary least squares (OLS) regression models that express tolls as functions of independent variables. Three families of models were estimated: linear models, models based on expansions of Taylor series, and models based on piece-wise linear approximations to non-linear effects. The resulting models were analyzed to identify the salient features of current toll policies towards different vehicle types.     

Implicit choice set generation in discrete choice models: Application to household auto ownership decisions

Latent choice set models that account for probabilistic consideration of choice alternatives during decision making have long existed. The Manski model that assumes a two-stage representation of decision making has served as the standard workhorse model for discrete choice modeling with latent choice sets. However, estimation of the Manski model is not always feasible because evaluation of the likelihood function in the Manski model requires enumeration of all possible choice sets leading to explosion for moderate and large choice sets. In this study, we propose a new group of implicit choice set generation models that can approximate the Manski model while retaining linear complexity with respect to the choice set size. We examined the performance of the models proposed in this study using synthetic data. The simulation results indicate that the approximations proposed in this study perform considerably well in terms of replicating the Manski model parameters. We subsequently used these implicit choice set models to understand latent choice set considerations in household auto ownership decisions of resident population in the Southern California region. The empirical results confirm our hypothesis that certain segments of households may only consider a subset of auto ownership levels while making decisions regarding the number of cars to own. The results not only underscore the importance of using latent choice models for modeling household auto ownership decisions but also demonstrate the applicability of the approximations proposed in this study to estimate these latent choice set models.     

Estimation of origin-destination matrices from link counts and sporadic routing data

Estimation of origin-destination (OD) matrices from link count data is a challenging problem because of the highly indeterminate relationship between the observations and the latent route flows. Conversely, estimation is straightforward if we observe the path taken by each vehicle. We consider an intermediate problem of increasing practical importance, in which link count data is supplemented by routing information for a fraction of vehicles on the network. We develop a statistical model for these combined data sources and derive some tractable normal approximations thereof. We examine likelihood-based inference for these normal models under the assumption that the probability of vehicle tracking is known. We show that the likelihood theory can be non-standard because of boundary effects, and provide conditions under which such irregular behaviour will be observed in practice. For regular cases we outline connections with existing generalised least squares methods. We then consider estimation of OD matrices under estimated and/or misspecified models for the probability of vehicle tracking. Theoretical developments are complemented by simulation experiments and an illustrative example using a section of road network from the English city of Leicester.     

Assessment of schedule-based and frequency-based assignment models for strategic and operational planning of high-speed rail services

Despite some substantial limitations in the simulation of low-frequency scheduled services, frequency-based (FB) assignment models are by far the most widely used in practice. They are less expensive to build and less demanding from the computational viewpoint with respect to schedule-based (SB) models, as they require neither explicit simulation of the timetable (on the supply side), nor segmentation of OD matrices by desired departure/arrival time (on the demand side).The objective of this paper is to assess to what extent the lack of modeling capabilities of FB models is acceptable, and, on the other hand, the cases in which such approximations are substantial and more detailed SB models are needed. This is a first attempt to shed light on the trade-off between (frequency-based) model inaccuracy and (scheduled-based) model development costs in the field of long-distance (e.g. High-speed Rail, HSR) service modeling.To this aim, we considered two modeling specifications estimated using mixed Revealed Preferences (RP) and Stated Preferences (SP) surveys and validated with respect to the same case study. Starting from an observed (baseline) scenario, we artificially altered the demand distributions (uniform vs. time-varying demand) and the supply configuration (i.e. train departure times), and analyzed the differences in modal split estimates and flows on individual trains, using the two different model specifications.It resulted that when the demand distribution is uniform within the period of analysis, such differences are significant only when departure times of trains are strongly unevenly spaced in time. In such cases, the difference in modal shares, using FB w.r.t. SB, is in the range of [0%, +5%] meaning that FB models tend to overestimate HSR modal shares. When the demand distribution is not uniform, the difference in modal shares, using FB w.r.t. SB, is in the range of [−10%, +10%] meaning that FB models can overestimate or underestimate HSR modal shares, depending on timetable settings with respect to travelers’ desired departure times. The differences in on-board train flow estimates are more substantial in both cases of uniform and not uniform demand distribution.     

An approximate analytic model of supertanker lightering operations

The most economical means of transporting crude oil over long distances is through the use of very large and ultra large crude containers (VLCCs and ULCCs respectively). These ships require sea lane depths of more than 55 feet to navigate. Since no major U.S. port is deeper than 45 feet, special offloading procedures must be employed. One such procedure is lightering, the process of offloading crude oil from VLCCs to smaller vessels for final delivery to the port.A linked set of queuing models is developed in an effort to understand lightering operations better and to enhance tradeoff analyses. The model assumes that port service times, lightering vessel loading times and VLCC interarrival times are exponentially distributed, allowing us to model lightering vessel operations using a cyclic queuing model. VLCC delays are modeled using an approximate M/E_k/S queue. The two models are linked through a VLCC service time model. The sensitivity of the model to changes in key inputs as well as the likely impacts of the model's assumptions and approximations are discussed. The paper concludes with recommendations for future studies.     

Improved models for technology choice in a transit corridor with fixed demand

We present three extensions to a base optimization model for a transit line which can be used to strategically evaluate technology choices. We add to the base model optimal stop spacing and train length, a crowding penalty, and a multi-period generalization. These extensions are analytically solvable by simple approximations and lead to meaningful insights. Their significance is illustrated by means of an example in which two road modes and two rail modes are defined by a set of techno-economical parameters. These parameters loaded in the base model yield dominance of road modes for all but the largest demand levels. We consistently keep this set of parameters for all models, and show how the break-even points between road and rail modes progressively recede toward lower demand levels when model refinements – not parameter changes – are applied. Scenario analyses of plausible parameter sets highlight the model’s versatility, and caution on general conclusions regarding technology dominance.     

Radial bus networks with multi-period and many-to-many demand

Mathematical models are developed for optimizing radial bus networks with time dependent demand and supply characteristics. These models can deal with many-to-many demand distribution in heterogeneous rather than idealized geographic environments. With some approximations, closed-form solutions for the optimal route angle, headways for different time periods, and stop spacings for various locations are obtained for a total cost minimization objective. The relations between the decision variables and system parameters are identified analytically. The optimality of a constant ratio between headways and route angle is found to hold with a time related factor. The optimized wait cost, operator cost, and lateral access cost are found to be equal. A numerical example is given for a case with three service periods. It illustrates the applicability of the analytic model to irregular demand patterns that may be directionally imbalanced during some periods.     

A review of vehicle fuel consumption models to evaluate eco-driving and eco-routing

Fuel consumption models have been widely used to predict fuel consumption and evaluate new vehicle technologies. However, due to the uncertainty and high nonlinearity of fuel systems, it is difficult to develop an accurate fuel consumption model for real-time calculations. Additionally, whether the developed fuel consumption models are suitable for eco-routing and eco-driving systems is unknown. To address these issues, a systematic review of fuel consumption models and the factors that influence fuel economy is presented. First, the primary factors that affect fuel economy, including travel-related, weather-related, vehicle-related, roadway-related, traffic-related, and driver-related factors, are discussed. Then, state-of-the-art fuel consumption models developed after 2000 are summarized and classified into three broad types based on transparency, i.e., white-box, grey-box and black-box models. Consequently, the limitations and potential possibilities of fuel consumption modelling are highlighted in this review.     

A novel variable selection method based on frequent pattern tree for real-time traffic accident risk prediction

With the availability of large volumes of real-time traffic flow data along with traffic accident information, there is a renewed interest in the development of models for the real-time prediction of traffic accident risk. One challenge, however, is that the available data are usually complex, noisy, and even misleading. This raises the question of how to select the most important explanatory variables to achieve an acceptable level of accuracy for real-time traffic accident risk prediction. To address this, the present paper proposes a novel Frequent Pattern tree (FP tree) based variable selection method. The method works by first identifying all the frequent patterns in the traffic accident dataset. Next, for each frequent pattern, we introduce a new metric, herein referred to as the Relative Object Purity Ratio (ROPR). The ROPR is then used to calculate the importance score of each explanatory variable which in turn can be used for ranking and selecting the variables that contribute most to explaining the accident patterns. To demonstrate the advantages of the proposed variable selection method, the study develops two traffic accident risk prediction models, based on accident data collected on interstate highway I-64 in Virginia, namely a k-nearest neighbor model and a Bayesian network. Prior to model development, two variable selection methods are utilized: (1) the FP tree based method proposed in this paper; and (2) the random forest method, a widely used variable selection method, which is used as the base case for comparison. The results show that the FP tree based accident risk prediction models perform better than the random forest based models, regardless of the type of prediction models (i.e. k-nearest neighbor or Bayesian network), the settings of their parameters, and the types of datasets used for model training and testing. The best model found is a FP tree based Bayesian network model that can predict 61.11% of accidents while having a false alarm rate of 38.16%. These results compare very favorably with other accident prediction models reported in the literature.     

A stochastic passenger loading model of airline schedule performance

The passenger loading problem is defined as that of determining the distribution of passengers that will be carried on each flight in a given market over the course of the day. This problem is stochastic in nature, and must take into account the fact that as some flights become booked up, passengers may spill to adjacent flights, or may choose not to fly on that airline. A precise model of the loading process is developed and solved using an efficient numerical procedure. Several approximations are introduced and tested which further improve the overall efficiency. The model represents a more rigorous solution approach than has appeared previously in the literature, and as such could be used to evaluate simpler approximations. It is fast enough, however, to be used as an interactive schedule evaluation and design tool.     

Evaluation of vehicle acceleration models for emission estimation at an intersection

We evaluate the constant acceleration, linearly decreasing acceleration, and aaSIDRA models in terms of generating second-by-second speed profiles for emission estimations at an intersection. The models are first calibrated using field data from individual vehicle trajectories. With the calibrated models, second-by-second speed and acceleration data are produced, and emissions are estimated using MOVES. Emission estimations based on the calibrated acceleration models are then compared with those based on field trajectory data. The constant acceleration model tends to overestimate emissions; both the linearly decreasing acceleration model and the aaSIDRA model provide accurate emission estimations.     

Online calibration of an integrated framework for information‐based evacuation operations

This study seeks to online calibrate the parameters of aggregate evacuee behavior models used in a behavior‐consistent information‐based control module for determining information strategies for real‐time evacuation operations. It enables the deployment of an operational framework for mass evacuation that integrates three aspects underlying an evacuation operation: demand (evacuee behavior), supply (network management), and disaster characteristics. To attain behavior‐consistency, the control module factors evacuees' likely responses to the disseminated information in determining information‐based control strategies. Hence, the ability of the behavior models to predict evacuees' likely responses is critical to the effectiveness of traffic routing by information strategies. The mixed logit structure is used for the aggregate behavior models to accommodate the behavioral heterogeneity across the population. An online calibration problem is proposed to calibrate the random parameters in the behavior models by using the least square estimator to minimize the gap between the predicted network flows and unfolding traffic dynamics. Background traffic, an important but rarely studied issue for modeling evacuation traffic, is also accounted for in the proposed problem. Numerical experiments are conducted to illustrate the importance of the calibration problem for addressing the system consistency issues and integrating the demand, supply, and disaster characteristics for more efficient evacuation operations. Copyright © 2016 John Wiley & Sons, Ltd.     

Vehicular pollution prediction modelling: a review of highway dispersion models

Air quality modelling plays an important role in formulating air pollution control and management strategies by providing guidelines for better and more efficient air quality planning. Several line source models, mostly Gaussian‐based, have been suggested to predict pollutant concentrations near highways/roads. These models, despite several assumptions and limitations, are used throughout the world, including in India, to carry out air pollution prediction analysis due to vehicular traffic near roads/highways. These models are being continuously upgraded and modified based on field experiments, and numerical and physical modelling results. An effort has been made in the present paper to review briefly the philosophy and basic features of most of the commonly used highway dispersion models. The paper also discusses various theories and techniques that led to the development and modification of these models along with the statistical analysis tools to evaluate the performance of these models. An attempt has also been made to summarize briefly the various line source models currently used in India and to highlight the difficulties being faced while using them in an Indian context.     

A model of daily time allocation to discretionary out-of-home activities and trips

This paper presents a model of discrete activity choice and continuous resource allocation which is based on the premise of random utility maximization and which can be conveniently estimated using existing statistical software packages. The model derivation involves virtually no approximations and adheres strictly to the utility maximization concept. The empirical analysis applies the model to the participation choice and resource (time) allocation to nonwork, out-of-home activities by workers. The statistical results show that activity choice and time allocation are governed by the same mechanism as the utilitarian assumptions indicate and support the theoretical framework employed in the model development.     

Evaluation of the General Motors based car-following models and a proposed fuzzy inference model

This paper evaluates the properties of the General Motors (GM) based car-following models, identifies their characteristics, and proposes a fuzzy inference logic based model that can overcome some of the shortcomings of the GM based models. This process involves developing a framework for evaluating a car-following model and comparing the behavior predicted by the GM models with the behavior observed under the real world situation. For this purpose, an instrumented vehicle was used to collect data on the headway and speeds of two consecutive vehicles under actual traffic conditions. Shortcomings of the existing GM based models are identified, in particular, the stability conditions were analyzed in detail. A fuzzy-inference based model of car-following is developed to represent the approximate nature of stimulus–response process during driving. This model is evaluated using the same evaluation framework used for the GM models and the data obtained by the instrumented test vehicle. Comparison between the performance of the two models show that the proposed fuzzy inference model can overcome many shortcomings of the GM based car-following models, and can be useful for developing the algorithm for the adaptive cruise control for automated highway system (AHS).     

Macroscopic traffic models for vehicle-follower automated transportation systems

The problem of distributing and routing vehicles in a large automated transportation network may be approached through the design of on-line control algorithms, particularly when the network contains many origin-destination pairs and alternate routes. To develop such algorithms, it is necessary to obtain models that accurately represent the dynamic behavior of vehicles on the guideway network. In this paper, models based on density, flow and average velocity variables are derived for the vehicle-follower longitudinal control scheme. Models suitable for use in analysis and simulation work are developed for links, merges, diverges, and stations. The proposed models are shown to compare favorably with simulation results that use explicit modeling of vehicle dynamic modeling of vehicle dynamic interaction.