Hubris or humility? Accuracy issues for the next 50 years of travel demand modeling

This study reviews the 50-year history of travel demand forecasting models, concentrating on their accuracy and relevance for public decision-making. Only a few studies of model accuracy have been performed, but they find that the likely inaccuracy in the 20-year forecast of major road projects is ±30 % at minimum, with some estimates as high as ±40–50 % over even shorter time horizons. There is a significant tendency to over-estimate traffic and underestimate costs, particularly for toll roads. Forecasts of transit costs and ridership are even more uncertain and also significantly optimistic. The greatest knowledge gap in US travel demand modeling is the unknown accuracy of US urban road traffic forecasts. Modeling weaknesses leading to these problems (non-behavioral content, inaccuracy of inputs and key assumptions, policy insensitivity, and excessive complexity) are identified. In addition, the institutional and political environments that encourage optimism bias and low risk assessment in forecasts are also reviewed. Major institutional factors, particularly low local funding matches and competitive grants, confound scenario modeling efforts and dampen the hope that technical modeling improvements alone can improve forecasting accuracy. The fundamental problems are not technical but institutional: high non-local funding shares for large projects warp local perceptions of project benefit versus costs, leading to both input errors and political pressure to fund projects. To deal with these issues, the paper outlines two different approaches. The first, termed ‘hubris’, proposes a multi-decade effort to substantially improve model forecasting accuracy over time by monitoring performance and improving data, methods and understanding of travel, but also by deliberately modifying the institutional arrangements that lead to optimism bias. The second, termed ‘humility’, proposes to openly quantify and recognize the inherent uncertainty in travel demand forecasts and deliberately reduce their influence on project decision-making. However to be successful either approach would require monitoring and reporting accuracy, standards for modeling and forecasting, greater model transparency, educational initiatives, coordinated research, strengthened ethics and reduction of non-local funding ratios so that localities have more at stake.     

Measuring inaccuracy in travel demand forecasting: methodological considerations regarding ramp up and sampling

Project promoters, forecasters, and managers sometimes object to two things in measuring inaccuracy in travel demand forecasting: (1) using the forecast made at the time of making the decision to build as the basis for measuring inaccuracy and (2) using traffic during the first year of operations as the basis for measurement. This paper presents the case against both objections. First, if one is interested in learning whether decisions about building transport infrastructure are based on reliable information, then it is exactly the traffic forecasted at the time of making the decision to build that is of interest. Second, although ideally studies should take into account so-called demand “ramp up” over a period of years, the empirical evidence and practical considerations do not support this ideal requirement, at least not for large-N studies. Finally, the paper argues that large samples of inaccuracy in travel demand forecasts are likely to be conservatively biased, i.e., accuracy in travel demand forecasts estimated from such samples would likely be higher than accuracy in travel demand forecasts in the project population. This bias must be taken into account when interpreting the results from statistical analyses of inaccuracy in travel demand forecasting.     

Traffic forecasts under uncertainty and capacity constraints

Traffic forecasts provide essential input for the appraisal of transport investment projects. However, according to recent empirical evidence, long-term predictions are subject to high levels of uncertainty. This article quantifies uncertainty in traffic forecasts for the tolled motorway network in Spain. Uncertainty is quantified in the form of a confidence interval for the traffic forecast that includes both model uncertainty and input uncertainty. We apply a stochastic simulation process based on bootstrapping techniques. Furthermore, the article proposes a new methodology to account for capacity constraints in long-term traffic forecasts. Specifically, we suggest a dynamic model in which the speed of adjustment is related to the ratio between the actual traffic flow and the maximum capacity of the motorway. As an illustrative example, this methodology is applied to a specific public policy that consists of suppressing the toll on a certain motorway section before the concession expires.     

Do Road Planners Produce More ‘Honest Numbers’ than Rail Planners? An Analysis of Accuracy in Road‐traffic Forecasts in Cities versus Peripheral Regions

Abstract

Based on a review of available data from a database on large‐scale transport infrastructure projects, this paper investigates the hypothesis that traffic forecasts for road links in Europe are geographically biased with underestimated traffic volumes in metropolitan areas and overestimated traffic volumes in remote regions. The present data do not support this hypothesis. Since previous studies have shown a strong tendency to overestimated forecasts of the number of passengers on new rail projects, it could be speculated that road planners are more skilful and/or honest than rail planners. However, during the period when the investigated projects were planned (up to the late 1980s), there were hardly any strong incentives for road planners to make biased forecasts in order to place their projects in a more flattering light. Future research might uncover whether the change from the ‘predict and provide’ paradigm to ‘predict and prevent’ occurring in some European countries in the 1990s has influenced the accuracy of road traffic forecasts in metropolitan areas.     

Selection bias in build-operate-transfer transportation project appraisals

Recent empirical studies have found widespread inaccuracies in traffic forecasts despite the fact that travel demand forecasting models have been significantly improved over the past few decades. We suspect that an intrinsic selection bias may exist in the competitive project appraisal process, in addition to the many other factors that contribute to inaccurate traffic forecasts. In this paper, we examine the potential for selection bias in the governmental process of Build-Operate-Transfer (BOT) transportation project appraisals. Although the simultaneous consideration of multiple criteria is typically used in practice, traffic flow estimate is usually a key criterion in these appraisals. For the purposes of this paper, we focus on the selection bias associated with the highest flow estimate criterion. We develop two approaches to quantify the level and chance of inaccuracy caused by selection bias: the expected value approach and the probability approach. The expected value approach addresses the question “to what extent is inaccuracy caused by selection bias?”. The probability approach addresses the question “what is the chance of inaccuracy due to selection bias?”. The results of this analysis confirm the existence of selection bias when a government uses the highest traffic forecast estimate as the priority criterion for BOT project selection. In addition, we offer some insights into the relationship between the extent/chance of inaccuracy and other related factors. We do not argue that selection bias is the only reason for inaccurate traffic forecasts in BOT projects; however, it does appear that it could be an intrinsic factor worthy of further attention and investigation.     

Determinants of air cargo traffic in California

Studies on the economic impacts of air cargo traffic have been gaining traction in recent years. The slowed growth of air cargo traffic at California’s airports, however, has raised pressing questions about the determinants of air cargo traffic. Specifically, it would be useful to know how California’s air cargo traffic is affected by urban economic characteristics. Accordingly, this study estimates the socioeconomic determinants of air cargo traffic across cities in California. We construct a 7-year panel (2003–2009) using quarterly employment, wage, population, and traffic data for metro areas in the state. Our results reveal that the concentrations of both service and manufacturing employment impact the volume of outbound air cargo. Total air cargo traffic is found to grow faster than population, while the corresponding domestic traffic grows less than proportionally to city size. Wages play a significant role in determining both total and domestic air cargo movement. We provide point estimates for traffic diversion between cities, showing that 80% of air cargo traffic is diverted away from a small city located within 100 miles of a large one. Using socioeconomic and demographic forecasts prepared for California’s Department of Transportation, we also forecast metro-level total and domestic air cargo tonnage for the years 2010–2040. Our forecasts for this period indicate that California’s total (domestic) air cargo traffic will increase at an average rate of 5.9% (4.4%) per year.     

The Effectiveness of Conceptual Airport Terminal Designs

Abstract

Aviation passenger traffic is forecast to grow significantly over the next decade and beyond. To accommodate this growth will require investment in airport infrastructure, including terminals. These buildings represent large, lumpy investments, so it is important to provide the capacity to accommodate the forecast traffic. However, this depends on at least two factors: the accuracy of the forecast of future demand, and the process of translating these forecasts into designs. Error in either factor can be potentially catastrophic financially. Translating forecasts into designs depends on ‘rules of thumb’ formulae that convert design hour flows into area requirements for each terminal facility. This paper examines the process of translating demand forecasts into conceptual terminal designs. The basic methods used are outlined, and how they affect the conceptual terminal design process are revealed. A model for conceptual terminal design is derived, presented and validated based on a sample of UK airports. It is shown that even if demand forecasts can be taken to be completely accurate, there can still be errors in terminal design and size resulting from the use of these ‘rules of thumb’.     

Analysis of peak and non‐peak traffic forecasts using combined models

Accurate and timely traffic forecasting is crucial to effective management of intelligent transportation systems (ITS). To predict travel time index (TTI) data, we select six baseline individual predictors as basic combination components. Applying the one‐step‐ahead out‐of‐sample forecasts, the paper proposes several linear combined forecasting techniques. States of traffic situations are classified into peak and non‐peak periods. Based on detailed data analyses, some practical guidance and comments are given in what situation a combined model is better than an individual model or other types of combined models. Indicating which model is more appropriate in each state, persuasive comparisons demonstrate that the combined procedures can significantly reduce forecast error rates. It reveals that the approaches are practically promising in the field. To the best of our knowledge, it is the first time to systematically investigate these approaches in peak and non‐peak traffic forecasts. The studies can provide a reference for optimal forecasting model selection in each period. Copyright © 2010 John Wiley & Sons, Ltd.     

Tollroads are only part of the overall trip: the error of our ways in past willingness to pay studies

With rare exception, actual tollroad traffic in many countries has failed to reproduce forecast traffic levels, regardless of whether the assessment is made after an initial year of operation or as long as 10 years after opening. Pundits have offered many reasons for this divergence, including optimism bias, strategic misrepresentation, the promise to equity investors of early returns on investment, errors in land use forecasts, and specific assumptions underlying the traffic assignment models used to develop traffic forecasts. One such assumption is the selection of a behaviourally meaningful value of travel time savings (VTTS) for use in a generalised cost or generalised time user benefit expression that is the main behavioural feature of the traffic assignment (route choice) model. Numerous empirical studies using stated choice experiments have designed choice sets of alternatives as if users choose a tolled route or a free route under the (implied) assumption that the tolled route is tolled for the entire trip. Reality is often very different, with a high incidence of use of a non-tolled road leading into and connecting out of a tolled link. In this paper we recognise this feature of route choice and redesign the stated choice experiment to account for it. Furthermore, this study is a follow up to a previous study undertaken before a new toll road was in place, and it benefits from real exposure to the new toll road. We find that the VTTS is noticeably reduced, and if the VTTS is a significant contributing influence on errors on traffic forecasts, then the lower estimates make sense behaviourally.     

Forecasting journey time distribution with consideration to abnormal traffic conditions

Travel time is an important index for managers to evaluate the performance of transportation systems and an intuitive measure for travelers to choose routes and departure times. An important part of the literature focuses on predicting instantaneous travel time under recurrent traffic conditions to disseminate traffic information. However, accurate travel time prediction is important for assessing the effects of abnormal traffic conditions and helping travelers make reliable travel decisions under such conditions. This study proposes an online travel time prediction model with emphasis on capturing the effects of anomalies. The model divides a path into short links. A Functional Principal Component Analysis (FPCA) framework is adopted to forecast link travel times based on historical data and real-time measurements. Furthermore, a probabilistic nested delay operator is used to calculate path travel time distributions. To ensure that the algorithm is fast enough for online applications, parallel computation architecture is introduced to overcome the computational burden of the FPCA. Finally, a rolling horizon structure is applied to online travel time prediction. Empirical results for Guangzhou Airport Expressway indicate that the proposed method can capture an abrupt change in traffic state and provide a promising and reliable travel time prediction at both the link and path levels. In the case where the original FPCA is modified for parallelization, accuracy and computational effort are evaluated and compared with those of the sequential algorithm. The proposed algorithm is found to require only a piece rather than a large set of traffic incident records.     

Accuracy of the Gothenburg congestion charges forecast

This paper explores the accuracy of the transport model forecast of the Gothenburg congestion charges, implemented in 2013. The design of the charging system implies that the path disutility cannot be computed as a sum of link attributes. The route choice model is therefore implemented as a hierarchical algorithm, applying a continuous value of travel time (VTT) distribution. The VTT distribution was estimated from stated choice (SC) data. However, based on experience of impact forecasting with a similar model and of impact outcome of congestion charges in Stockholm, the estimated VTT distribution had to be stretched to the right. We find that the forecast traffic reductions across the cordon and travel time gains were close to those observed in the peak. However, the reduction in traffic across the cordon was underpredicted off-peak. The necessity to make the adjustment indicates that the VTT inferred from SC data does not reveal the travellers’ preferences, or that there are factors determining route choice other than those included in the model: travel distance, travel time and congestion charge.     

Trust in forecasts? Correlates with ridership forecast accuracy for fixed-guideway transit projects

Transportation - The accuracy of ridership forecasts for fixed-guideway transit projects in the United States has improved in recent decades. A better understanding of the causes for this...     

Validating the Passenger Traffic Model for Copenhagen

The paper presents a comprehensive validation procedure for the passenger traffic model for Copenhagen based on external data from the Danish national travel survey and traffic counts. The model was validated for the years 2000–2004, with 2004 being of particular interest because the Copenhagen Metro became operational in autumn 2002. We observed that forecasts from the demand sub-models agree well with the data from the 2000 national travel survey, with the mode choice forecasts in particular being a good match with the observed modal split. The results of the 2000 car assignment model matched the observed traffic better than those of the transit assignment model. With respect to the metro forecasts, the model over-predicts metro passenger flows by 10–50%. The wide range of findings from the project resulted in two actions. First, a project was started in January 2005 to upgrade the model’s base trip matrices. Second, a dialog between researchers and the Ministry of Transport has been initiated to discuss the need to upgrade the Copenhagen model, e.g. a switching to an activity-based paradigm and improving assignment procedures.     

Short-term traffic flow rate forecasting based on identifying similar traffic patterns

The ability to timely and accurately forecast the evolution of traffic is very important in traffic management and control applications. This paper proposes a non-parametric and data-driven methodology for short-term traffic forecasting based on identifying similar traffic patterns using an enhanced K-nearest neighbor (K-NN) algorithm. Weighted Euclidean distance, which gives more weight to recent measurements, is used as a similarity measure for K-NN. Moreover, winsorization of the neighbors is implemented to dampen the effects of dominant candidates, and rank exponent is used to aggregate the candidate values. Robustness of the proposed method is demonstrated by implementing it on large datasets collected from different regions and by comparing it with advanced time series models, such as SARIMA and adaptive Kalman Filter models proposed by others. It is demonstrated that the proposed method reduces the mean absolute percent error by more than 25%. In addition, the effectiveness of the proposed enhanced K-NN algorithm is evaluated for multiple forecast steps and also its performance is tested under data with missing values. This research provides strong evidence suggesting that the proposed non-parametric and data-driven approach for short-term traffic forecasting provides promising results. Given the simplicity, accuracy, and robustness of the proposed approach, it can be easily incorporated with real-time traffic control for proactive freeway traffic management.     

Travel demand forecasts improved by using cross-sectional data from multiple time points

Forecasts of travel demand are often based on data from the most recent time point, even when cross-sectional data is available from multiple time points. This is because forecasting models with similar contexts have higher transferability, and the context of the most recent time point is believed to be the most similar to the context of a future time point. In this paper, the author proposes a method for improving the forecasting performance of disaggregate travel demand models by utilising not only the most recent dataset but also an older dataset. The author assumes that the parameters are functions of time, which means that future parameter values can be forecast. These forecast parameters are then used for travel demand forecasting. This paper describes a case study of journeys to work mode choice analysis in Nagoya, Japan, using data collected in 1971, 1981, 1991, and 2001. Behaviours in 2001 are forecast using a model with only the most recent 1991 dataset and models that combine the 1971, 1981, and 1991 datasets. The models proposed by the author using data from three time points can provide better forecasts. This paper also discusses the functional forms for expressing parameter changes and questions the temporal transferability of not only alternative-specific constants but also level-of-service and socio-economic parameters.     

Impact of heavy vehicles on surrounding traffic characteristics

This work examines the impact of heavy vehicle movements on measured traffic characteristics in detail. Although the number of heavy vehicles within the traffic stream is only a small percentage, their impact is prominent. Heavy vehicles impose physical and psychological effects on surrounding traffic flow because of their length and size (physical) and acceleration/deceleration (operational) characteristics. The objective of this work is to investigate the differences in traffic characteristics in the vicinity of heavy vehicles and passenger cars. The analysis focuses on heavy traffic conditions (level of service E) using a trajectory data of highway I‐80 in California. The results show that larger front and rear space gaps exist for heavy vehicles compared with passenger cars. This may be because of the limitations in manoeuvrability of heavy vehicles and the safety concerns of the rear vehicle drivers, respectively. In addition, heavy vehicle drivers mainly keep a constant speed and do not change their speed frequently. This work also examines the impact of heavy vehicles on their surrounding traffic in terms of average travel time and number of lane changing manoeuvres using Advanced Interactive Microscopic Simulator for Urban and Non‐Urban Networks (AIMSUN) microscopic traffic simulation package. According to the results, the average travel time increases when proportion of heavy vehicles rises in each lane. To reflect the impact of heavy vehicles on average travel time, a term related to heavy vehicle percentage is introduced into two different travel time equations, Bureau of Public Roads and Akçelik's travel time equations. The results show that using an exclusive term for heavy vehicles can better estimate the travel times for more than 10%. Finally, number of passenger car lane changing manoeuvres per lane will be more frequent when more heavy vehicles exist in that lane. The influence of heavy vehicles on the number of passenger car lane changing is intensified in higher traffic densities and higher percentage of heavy vehicles. Large numbers of lane changing manoeuvres can increase the number of traffic accidents and potentially reduce traffic safety. The results show an increase of 5% in the likelihood of accidents, when percentage of heavy vehicles increases to 30% of total traffic. Copyright © 2014 John Wiley & Sons, Ltd.     

Roadway determinants of bicyclist exposure to volatile organic compounds and carbon monoxide

Few studies have quantified relationships between bicyclist exposure to air pollution and roadway and traffic variables. As a result, transportation professionals are unable to easily estimate exposure differences among bicycle routes for network planning, design, and analysis. This paper estimates the effects of roadway and travel characteristics on bicyclist exposure concentrations, controlling for meteorology and background conditions. Concentrations of volatile organic compounds (VOC) and carbon monoxide (CO) are modeled using high-resolution data collected on-road. Results indicate that average daily traffic (ADT) provides a parsimonious way to characterize the impact of roadway characteristics on bicyclists’ exposure. VOC and CO exposure increase by approximately 2% per 1000 ADT, robust to different regression model specifications. Exposure on off-street facilities is higher than at a park, but lower than on-street riding – with the exception of a path through an industrial corridor with significantly higher exposure. VOC exposure is 20% higher near intersections. Traffic, roadway, and travel variables have more explanatory power in the VOC models than the CO model. The quantifications in this paper enable calculation of expected exposure differences among travel paths for planning and routing applications. The findings also have policy and design implications to reduce bicyclists’ exposure. Separation between bicyclists and motor vehicle traffic is a necessary but not sufficient condition to reduce exposure concentrations; off-street paths are not always low-exposure facilities.     

Predicting travel time reliability using mobile phone GPS data

Estimates of road speeds have become commonplace and central to route planning, but few systems in production provide information about the reliability of the prediction. Probabilistic forecasts of travel time capture reliability and can be used for risk-averse routing, for reporting travel time reliability to a user, or as a component of fleet vehicle decision-support systems. Many of these uses (such as those for mapping services like Bing or Google Maps) require predictions for routes in the road network, at arbitrary times; the highest-volume source of data for this purpose is GPS data from mobile phones. We introduce a method (TRIP) to predict the probability distribution of travel time on an arbitrary route in a road network at an arbitrary time, using GPS data from mobile phones or other probe vehicles. TRIP captures weekly cycles in congestion levels, gives informed predictions for parts of the road network with little data, and is computationally efficient, even for very large road networks and datasets. We apply TRIP to predict travel time on the road network of the Seattle metropolitan region, based on large volumes of GPS data from Windows phones. TRIP provides improved interval predictions (forecast ranges for travel time) relative to Microsoft’s engine for travel time prediction as used in Bing Maps. It also provides deterministic predictions that are as accurate as Bing Maps predictions, despite using fewer explanatory variables, and differing from the observed travel times by only 10.1% on average over 35,190 test trips. To our knowledge TRIP is the first method to provide accurate predictions of travel time reliability for complete, large-scale road networks.     

Short‐term travel time forecasts for transport information system in hong kong

In this paper, a case study is carried out in Hong Kong for demonstration of the Transport Information System (TIS) prototype. A traffic flow simulator (TFS) is presented to forecast the short‐term travel times that can be served as a predicted travel time database for the TIS in Hong Kong. In the TFS, a stochastic deviation coefficient is incorporated to simulate the minute‐by‐minute fluctuation of traffic flows within the peak hour period. The purposes of the case study are: 1) to show the applicability of the TFS for larger‐scale road network; and 2) to illustrate the short‐term forecasting of path travel times in practice. The results of the case study show that the TFS can be applied to real network effectively. The predicted travel times are compared with the observed travel times on the selected paths for an OD pair. The results show that the observed path travel times fall in the 90% confidence interval of the predicted path travel times.     

Practical approach for travel time estimation from point traffic detector data

Accurate estimation of travel time is critical to the success of advanced traffic management systems and advanced traveler information systems. Travel time estimation also provides basic data support for travel time reliability research, which is being recognized as an important performance measure of the transportation system. This paper investigates a number of methods to address the three major issues associated with travel time estimation from point traffic detector data: data filling for missing or error data, speed transformation from time‐mean speed to space‐mean speed, and travel time estimation that converts the speeds recorded at detector locations to travel time along the highway segment. The case study results show that the spatial and temporal interpolation of missing data and the transformation to space‐mean speed improve the accuracy of the estimates of travel time. The results also indicate that the piecewise constant‐acceleration‐based method developed in this study and the average speed method produce better results than the other three methods proposed in previous studies. Copyright © 2011 John Wiley & Sons, Ltd.