On the imputation of missing data for road traffic forecasting: New insights and novel techniques

Vehicle flow forecasting is of crucial importance for the management of road traffic in complex urban networks, as well as a useful input for route planning algorithms. In general traffic predictive models rely on data gathered by different types of sensors placed on roads, which occasionally produce faulty readings due to several causes, such as malfunctioning hardware or transmission errors. Filling in those gaps is relevant for constructing accurate forecasting models, a task which is engaged by diverse strategies, from a simple null value imputation to complex spatio-temporal context imputation models. This work elaborates on two machine learning approaches to update missing data with no gap length restrictions: a spatial context sensing model based on the information provided by surrounding sensors, and an automated clustering analysis tool that seeks optimal pattern clusters in order to impute values. Their performance is assessed and compared to other common techniques and different missing data generation models over real data captured from the city of Madrid (Spain). The newly presented methods are found to be fairly superior when portions of missing data are large or very abundant, as occurs in most practical cases.     

Time-varying effects of influential factors on incident clearance time using a non-proportional hazard-based model

Incident clearance time is a major performance measure of the traffic emergency management. A clear understanding of the contributing factors and their effects on incident clearance time is essential for optimal incident management resource allocations. Most previous studies simply considered the average effects of the influential factors. Although the time-varying effects are also important for incident management agencies, they were not sufficiently investigated. To fill up the gap, this study develops a non-proportional hazard-based duration model for analyzing the time-varying effects of influential factors on incident clearance time. This study follows a systematic approach incorporating the following three procedures: proportionality test, model development/estimation, and effectiveness test. Applying the proposed model to the 2009 Washington State Incident Tracking System data, five factors were found to have significant but constant (or time independent) effects on the clearance time, which is similar to the findings from previous studies. However, our model also discovered thirteen variables that have significant time-varying impacts on clearance hazard. These factors cannot be identified through the conventional methods used in most previous studies. The influential factors are investigated from both macroscopic and microscopic perspectives. The population average effect evaluation provides the macroscopic insight and benefits long-term incident management, and the time-dependent pattern identification offers microscopic and time-sequential insight and benefits the specific incident clearance process.     

Overview of missing physical commodity trade data and its imputation using data augmentation

The physical aspects of commodity trade are becoming increasingly important on a global scale for transportation planning, demand management for transportation facilities and services, energy use, and environmental concerns. Such aspects (for example, weight and volume) of commodities are vital for logistics industry to allow for medium-to-long term planning at the strategic level and identify commodity flow trends. However, incomplete physical commodity trade databases impede proper analysis of trade flow between various countries. The missing physical values could be due to many reasons such as, (1) non-compliance of reporter countries with the prescribed regulations by World Customs Organization (WCO) (2) confidentiality issues, (3) delays in processing of data, or (4) erroneous reporting. The traditional missing data imputation methods, such as the substitution by mean, substitution by linear interpolation/extrapolation using adjacent points, the substitution by regression, and the substitution by stochastic regression, have been proposed in the context of estimating physical aspects of commodity trade data. However, a major demerit of these single imputation methods is their failure to incorporate uncertainty associated with missing data. The use of computationally complex stochastic methods to improve the accuracy of imputed data has recently become possible with the advancement of computer technology. Therefore, this study proposes a sophisticated data augmentation algorithm in order to impute missing physical commodity trade data. The key advantage of the proposed approach lies in the fact that instead of using a point estimate as the imputed value, it simulates a distribution of missing data through multiple imputations to reflect uncertainty and to maintain variability in the data. This approach also provides the flexibility to include fundamental distributional property of the variables, such as physical quantity, monetary value, price elasticity of demand, price variation, and product differentiation, and their correlations to generate reasonable average estimates of statistical inferences. An overview and limitations of most commonly used data imputation approaches is presented, followed by the theoretical basis and imputation procedure of the proposed approach. Lastly, a case study is presented to demonstrate the merits of the proposed approach in comparison to traditional imputation methods.     

An efficient realization of deep learning for traffic data imputation

Traffic data provide the basis for both research and applications in transportation control, management, and evaluation, but real-world traffic data collected from loop detectors or other sensors often contain corrupted or missing data points which need to be imputed for traffic analysis. For this end, here we propose a deep learning model named denoising stacked autoencoders for traffic data imputation. We tested and evaluated the model performance with consideration of both temporal and spatial factors. Through these experiments and evaluation results, we developed an algorithm for efficient realization of deep learning for traffic data imputation by training the model hierarchically using the full set of data from all vehicle detector stations. Using data provided by Caltrans PeMS, we have shown that the mean absolute error of the proposed realization is under 10 veh/5-min, a better performance compared with other popular models: the history model, ARIMA model and BP neural network model. We further investigated why the deep leaning model works well for traffic data imputation by visualizing the features extracted by the first hidden layer. Clearly, this work has demonstrated the effectiveness as well as efficiency of deep learning in the field of traffic data imputation and analysis.     

Heterogeneous hazard model of PEV users charging intervals: Analysis of four year charging transactions data

Public charging infrastructure represents a key success factor in the promotion of plug-in electric vehicles (PEV). Given that a large initial investment is required for the widespread adoption of PEV, many studies have addressed the location choice problem for charging infrastructure using a priori simple assumptions. Ideally, however, identifying optimal locations of charging stations necessitates an understanding of charging behavior. Limited market penetration of PEV makes it difficult to grasp any regularities in charging behavior. Using a Dutch data set about four-years of charging transactions, this study presents a detailed analysis of inter-charging times. Recognizing that PEV users may exhibit different charging behavior, this study estimates a latent class hazard duration model, which accommodates duration dependence, unobserved heterogeneity and the effects of time-varying covariates. PEV users are endogenously classified into regular and random users by treating charging regularity as a latent variable. The paper provides valuable insights into the dynamics of charging behavior at public charging stations, and which strategies can be successfully used to improve the performance of public charging infrastructure.     

An analysis of the effects of French vehicle inspection program and grant for scrappage on household vehicle transaction

The French government has implemented a periodical vehicle inspection program, which aims at maintaining proper functioning of the vehicle and ensuring the emissions control systems installed on the vehicle work properly. Also, an incentive program for scrapping old vehicles was introduced in 1994 through 1996 to promote the replacement of those vehicles with higher emissions by newer vehicles with lower emissions. A hazard-based duration model of household vehicle transaction behavior has been developed in this study to examine the effects of the inspection program and the grant for scrappage on vehicle transaction timing. The model is developed as a competing risks model assuming the following three types of competing risks: replacing one of the vehicles in the household fleet, disposing of one vehicle in the fleet, and acquiring one vehicle to add to the fleet. The empirical analysis is carried out using the panel data of French households' vehicle ownership from 1984 to 1998, obtained by the panel survey called Parc-Auto, which has been conducted by a French marketing firm, SOFRES, since 1976. The long panel observation period facilitates the introduction of macro-economic indicators into the model, enabling the analysis to distinguish the effects of policy measures from macro-economic factors. The empirical results indicate that the expected vehicle holding duration becomes 1.3 years longer under the inspection program than before the program commenced, given that the vehicle is replaced by another vehicle at the end of the holding duration; and that the conditional probability of replacing a vehicle aged 10 years and over becomes 1.2 times higher, and the average holding duration becomes shorter by 3.3 years, when the grant for scrappage is available.     

Competing risk mixture model and text analysis for sequential incident duration prediction

Predicting the duration of traffic incidents sequentially during the incident clearance period is helpful in deploying efficient measures and minimizing traffic congestion related to such incidents. This study proposes a competing risk mixture hazard-based model to analyze the effect of various factors on traffic incident duration and predict the duration sequentially. First, topic modeling, a text analysis technique, is used to process the textual features of the traffic incident to extract time-dependent topics. Given four specific clearance methods and the uncertainty of these methods when used during traffic incidents, the proposed mixture model uses the multinomial logistic model and parametric hazard-based model to assess the influence of covariates on the probability of clearance methods and on the duration of the incident. Subsequently, the performance of estimated mixture model in sequentially predicting the incident duration is compared with that of the non-mixture model. The prediction results show that the presented mixture model outperforms the non-mixture model.     

Comparison of advanced imputation algorithms for detection of transportation mode and activity episode using GPS data

Global Positioning System (GPS) technologies have been increasingly considered as an alternative to traditional travel survey methods to collect activity-travel data. Algorithms applied to extract activity-travel patterns vary from informal ad-hoc decision rules to advanced machine learning methods and have different accuracy. This paper systematically compares the relative performance of different algorithms for the detection of transportation modes and activity episodes. In particular, naive Bayesian, Bayesian network, logistic regression, multilayer perceptron, support vector machine, decision table, and C4.5 algorithms are selected and compared for the same data according to their overall error rates and hit ratios. Results show that the Bayesian network has a better performance than the other algorithms in terms of the percentage correctly identified instances and Kappa values for both the training data and test data, in the sense that the Bayesian network is relatively efficient and generalizable in the context of GPS data imputation.     

Driver's braking behavior approaching pedestrian crossings: a parametric duration model of the speed reduction times

The driver's braking behavior while approaching zebra crossings under different safety measures (curb extensions, parking restrictions, and advance yield markings) and without treatment (baseline condition) was examined. The speed reduction time was the variable used to describe the driver's behavior. Forty‐two drivers drove a driving simulator on an urban scenario in which the baseline condition and the safety measures were implemented. The speed reduction time was modeled with a parametric duration model to compare the effects on driver's braking behavior of vehicle dynamic variables and different countermeasures. The parametric accelerated failure time duration model with a Weibull distribution identified that the vehicle dynamic variables and only the countermeasure curb extensions affected, in a statistically significant way, the driver's speed reduction time in response to a pedestrian crossing. This result shows that the driver, because of the improved visibility of the pedestrian allowed by the curb extensions, was able to receive a clear information and better to adapt his approaching speed to yield to the pedestrian, avoiding abrupt maneuvers. This also means a reduction of likelihood of rear‐end collision due to less aggressive braking. Copyright © 2016 John Wiley & Sons, Ltd.     

A hybrid approach to integrate fuzzy C-means based imputation method with genetic algorithm for missing traffic volume data estimation

Although various innovative traffic sensing technologies have been widely employed, incomplete sensor data is one of the most major problems to significantly degrade traffic data quality and integrity. In this study, a hybrid approach integrating the Fuzzy C-Means (FCM)-based imputation method with the Genetic Algorithm (GA) is develop for missing traffic volume data estimation based on inductance loop detector outputs. By utilizing the weekly similarity among data, the conventional vector-based data structure is firstly transformed into the matrix-based data pattern. Then, the GA is applied to optimize the membership functions and centroids in the FCM model. The experimental tests are conducted to verify the effectiveness of the proposed approach. The traffic volume data collected at different temporal scales were used as the testing dataset, and three different indicators, including root mean square error, correlation coefficient, and relative accuracy, are utilized to quantify the imputation performance compared with some conventional methods (Historical method, Double Exponential Smoothing, and Autoregressive Integrated Moving Average model). The results show the proposed approach outperforms the conventional methods under prevailing traffic conditions.     

The net incidence of transit subsidies: a case study

The purpose of this paper is to investigate the net incidence of government subsidies to a transit system (i.e., the net impact of who pays and who benefits from transit subsidies). Rather than considering the U.S. transit subsidy program in the aggregate, the net incidence of subsidies to a particular transit system — Tidewater Regional Transit (TRT) — is analyzed. The paper concludes that the net incidence of the TRT subsidy program is progressive. Furthermore, the paper provides a methodology that can be used for investigating the net incidence of government subsidies to other transit systems.     

Use of duration models for predicting vehicular delay at a US/Canadian border crossing

Delays caused by congestion at the US/Canadian border crossing between Washington state and British Columbia have underscored the need for some sort of intervention. One obvious congestion-mitigation measure would be to estimate delay times and then relay this information to motorists so that they could select among alternative border crossing sites, or delay their trips. This paper applies duration models to estimate vehicular delay and demonstrates the usefulness of such models as a basis for a fully automated motorist information system. The paper also explores the flexibility of duration models, in providing estimates of vehicle delay, by using alternate parametric forms and assessing prediction accuracy.     

A comparative study on data segregation for mesoscopic energy modeling

On-road vehicles have been considered as one of the major contributors to energy consumption and air pollutant emissions. In order to quantify the corresponding environmental impacts, great efforts have been dedicated to the microscopic and macroscopic modeling for vehicle energy consumption and emissions. However, the mesoscopic modeling research that is focused on estimating trip-based energy consumption and is critical to some ITS applications (e.g., environmentally-friendly navigation), is relatively deficient. This study aims to investigate the effects of different data segregation methods on the mesoscopic modeling for vehicle energy consumption. A variety of novel methods, including the so-called conditional operating mode based method, have been proposed and evaluated using field data. Based on real-world data, statistical analyses have demonstrated the superior performance of enhanced models (i.e., conditional operating mode/VSP based models) in estimating vehicle energy consumption on a trip basis, compared to the other four models (velocity binning, time snipping, distance snipping and VSP based models) tested in this study.     

A multinomial logit car use model for a megacity of the developing world: Istanbul

The increase in motor vehicle use is one of the important factors that cause traffic congestion, especially in megacities. Thus, the reasons behind this increase require serious attention. This paper offers an analysis of this kind, for a megacity from the developing world, Istanbul. A stratified multinomial logit model accounting for the availability of a second vehicle in the household is estimated for a sample drawn from a questionnaire to gather information of actual car use in Istanbul. This estimation is only possible through a unique data generation process that converts actual preferences into a choice study setting. In addition, a simulation study, generally utilized in the analyses of discrimination between certain layers of society, and a scenario analysis related to changes in income are also included in the paper for a better understanding of the nature of the topic. The results show that the behavior of households with a second vehicle available and not available varies significantly due to household, individual and professional-related characteristics.     

Examination of factors affecting freeway incident clearance times: a comparison of the generalized F model and several alternative nested models

Traffic incidents are a principal cause of congestion on urban freeways, reducing capacity and creating risks for both involved motorists and incident response personnel. As incident durations increase, the risk of secondary incidents or crashes also becomes problematic. In response to these issues, many road agencies in metropolitan areas have initiated incident management programs aimed at detecting, responding to, and clearing incidents to restore freeways to full capacity as quickly and safely as possible. This study examined those factors that impact the time required by the Michigan Department of Transportation Freeway Courtesy Patrol to clear incidents that occurred on the southeastern Michigan freeway network. These models were developed using traffic flow data, roadway geometry information, and an extensive incident inventory database. A series of parametric hazard duration models were developed, each assuming a different underlying probability distribution for the hazard function. Although each modeling framework provided results that were similar in terms of the direction of factor effects, there was significant variability in terms of the estimated magnitude of these impacts. The generalized F distribution was shown to provide the best fit to the incident clearance time data, and the use of poorer fitting distributions was shown to result in severe over‐estimation or under‐estimation of factor effects. Those factors that were found to impact incident clearance times included the time of day and month when the incident occurred, the geometric and traffic characteristics of the freeway segment, and the characteristics of each incident. Copyright © 2012 John Wiley & Sons, Ltd.     

Estimation of a route choice model for urban public transport using smart card data

This paper describes a logit model of route choice for urban public transport and explains how the archived data from a smart card-based fare payment system can be used for the choice set generation and model estimation. It demonstrates the feasibility and simplicity of applying a trip-chaining method to infer passenger journeys from smart card transactions data. Not only origins and destinations of passenger journeys can be inferred but also the interchanges between the segments of a linked journey can be recognised. The attributes of the corresponding routes, such as in-vehicle travel time, transfer walking time and to get from alighting stop to trip destination, the need to change, and the time headway of the first transportation line, can be determined by the combination of smart card data with other data sources, such as a street map and timetable. The smart card data represent a large volume of revealed preference data that allows travellers' behaviour to be modelled with higher accuracy than by using traditional survey data. A multinomial route choice model is proposed and estimated by the maximum likelihood method, using urban public transport in ?ilina, the Slovak Republic, as a case study     

A data perception model for the safe operation of high-speed rail in rainstorms

In order to solve the safety operation problems of High-speed rail (HSR) in different areas and different sections under the rainstorm condition, an early warning process for the rainstorm disaster is designed. Furthermore, in order to control the operation risk, a HSR operation program with different rainstorm degrees is given out based on the analysis of rainstorm warning mechanism and rainstorm warning threshold in this paper. In addition, considering the reality that natural conditions vary greatly and the rainfall is very uneven, a data perception model of rainstorm (DPM) is proposed with correction coefficients for solving the calculation problem of precipitation for rainstorm warning. The DPM mainly adopts Paulhus’s empirical equation and uses the linear function to improve it for calculating the precipitation, which is able to calculate the hourly precipitation in different regional environments, and also effectively evaluate the rainstorm warning level of high-speed rail in this period. It can calculate and monitor the process by big data and MATLAB. The result of case analysis shows that the DPM has good practical value for solving the safety operation problem of HSR in different areas under rainstorm environment.     

Efficiency of a highway use reservation system for morning commute

This paper examines the design and efficiency of a highway use reservation system where commuters need reservations to access a highway facility at specific times. We show that, by accommodating reservation requests to the level that the highway capacity allows, traffic congestion can be relieved. Generally, a more differentiated design of the reservation system yields a higher reduction of travel cost and thus achieves a higher efficiency. The efficiency bound of the system is established. We also show that braking or tactical waiting behaviors of drivers would cause a loss of efficiency, which thus need be proactively accommodated. Given that user heterogeneity cause further loss of efficiency, we explore how two specific types of user heterogeneity affect the system efficiency. Auction-based reservation is then proposed to mitigate the efficiency loss.     

The reliability of using the gravity model for forecasting trip distribution

The main purpose of this study is to assess the forecasting capability of the gravity model and to investigate the merit of including K-factors when using the model. Peak hour trip data was obtained for four study year periods 1962, 1971, 1976 and 1981 for the City of Winnipeg. Analysis of the calibration results indicated that the F-factors for the twenty year period were stable within a range of values. In general, however, the K-factors were found to be inconsistent from one prediction period to the next, and when used in forecasting trips they resulted in larger errors than without their use. The validity of using K-factors or the method which has been used to determine them is questionable. It was concluded that while K-factors are very meaningful in theory (as defined), they are not appropriate for use in predicting O-D matrices based on the method by which they are currently estimated (i.e. as a simple ratio). Further study is needed to investigate an alternative method of calibrating the gravity model such as the cell-by-cell regression method.     

Development of a delay model for unsignalized intersections applicable to traffic assignment

Abstract

This paper develops a model for estimating unsignalized intersection delays which can be applied to traffic assignment (TA) models. Current unsignalized intersection delay models have been developed mostly for operational purposes, and demand detailed geometric data and complicated procedures to estimate delay. These difficulties result in unsignalized intersection delays being ignored or assumed as a constant in TA models.

Video and vehicle license plate number recognition methods are used to collect traffic volume data and to measure delays during peak and off-peak traffic periods at four unsignalized intersections in the city of Tehran, Iran. Data on geometric design elements are measured through field surveys. An empirical approach is used to develop a delay model as a function of influencing factors based on 5- and 15-min time intervals. The proposed model estimates delays on each approach based on total traffic volumes, rights-of-way of the subject approach and the intersection friction factor. The effect of conflicting traffic flows is considered implicitly by using the intersection friction factor. As a result, the developed delay model guarantees the convergence of TA solution methods.

A comparison between delay models performed using different time intervals shows that the coefficients of determination, R ², increases from 43.2% to 63.1% as the time interval increases from 5- to 15-min. The US Highway Capacity Manual (HCM) delay model (which is widely used in Iran) is validated using the field data and it is found that it overestimates delay, especially in the high delay ranges.