Eleni I. Vlahogianni

Short‐term traffic forecasting: Overview of objectives and methods

In the last two decades, the growing need for short‐term prediction of traffic parameters embedded in a real‐time intelligent transportation systems environment has led to the development of a vast number of forecasting algorithms. Despite this, there is still not a clear view about the various requirements involved in modelling. This field of research was examined by disaggregating the process of developing short‐term traffic forecasting algorithms into three essential clusters: the determination of the scope, the conceptual process of specifying the output and the process of modelling, which includes several decisions concerning the selection of the proper methodological approach, the type of input and output data used, and the quality of the data. A critical discussion clarifies several interactions between the above and results in a logical flow that can be used as a framework for developing short‐term traffic forecasting models.

Spatio-Temporal Short-Term Urban Traffic Volume Forecasting Using Genetically Optimized Modular Networks

Current interest in short-term traffic volume forecasting focuses on incorporating temporal and spatial volume characteristics in the forecasting process. This paper addresses the problem of integrating and optimizing predictive information from multiple locations of an urban signalized arterial roadway and proposes a modular neural predictor consisting of temporal genetically optimized structures of multilayer perceptrons that are fed with volume data from sequential locations to improve the accuracy of short-term forecasts. Results show that the proposed methodology provides more accurate forecasts compared to the conventional statistical methodologies applied, as well as to the static forms of neural networks.

Temporal Evolution of Short‐Term Urban Traffic Flow: A Nonlinear Dynamics Approach

Recognizing temporal patterns in traffic flow has been an important consideration in short- term traffic forecasting research. However, little work has been conducted on identifying and associating traffic pattern occurrence with prevailing traffic con- ditions. We propose a multilayer strategy that first identifies patterns of traffic based on their structure and evolution in time and then clusters the pattern- based evolution of traffic flow with respect to pre- vailing traffic flow conditions. Temporal pattern iden- tification is based on the statistical treatment of the recurrent behavior of jointly considered volume and oc- cupancy series; clustering is done via a two-level neural network approach. Results on urban signalized arterial 90-second traffic volume and occupancy data indicate that traffic pattern propagation exhibits variability with respect to its statistical characteristics such as determinis- tic structure and nonlinear evolution. Further, traffic pat- tern clustering uncovers four distinct classes of traffic pat- tern evolution, whereas transitional traffic conditions can be straightforwardly identified.

Modeling the Effects of Weather and Traffic on the Risk of Secondary Incidents

To extract useful information on variables that are associated with secondary accident likelihood, this article develops neural network models with enhanced explanatory power. Traffic and weather conditions at the occurrence of a primary incident are explicitly considered. Two measures to extract variable significance are introduced: mutual information and partial derivatives. The proposed approach is also compared to other classical statistical approaches of the Logit family. Results suggest that traffic speed, duration of the primary accident, hourly volume, rainfall intensity, and number of vehicles involved in the primary accident are the top five factors associated with secondary accident likelihood. However, changes in traffic speed and volume, number of vehicles involved, blocked lanes, and percentage of trucks and upstream geometry also significantly influence the probability of having a secondary incident. Finally, the incident management implications of the proposed modeling approach are discussed.

Freeway Operations, Spatiotemporal-Incident Characteristics, and Secondary-Crash Occurrence

Incidents are a major source of uncertainty in freeway operations. Secondary crashes are an important category of freeway incidents. Until now, secondary crashes have been assumed to occur at the boundary of high-density queues formed upstream of an initial crash. While much research has concentrated on the relationship between incident duration and secondary-crash likelihood, the incidents influence area is widely treated as independent of prevailing traffic conditions and incident characteristics. This paper extends research by developing a Bayesian network for the probabilistic estimation of different influence areas for secondary-crash occurrence with respect to various incident and traffic characteristics. Results indicate that traffic conditions at the time of an incident, as well as the time needed to respond to and clear the crash scene, are the most significant determinants in defining the upstream influence area of a crash.

Fuzzy-Entropy Neural Network Freeway Incident Duration Modeling with Single and Competing Uncertainties

This article describes an approach for predicting incident durations that are susceptible to severe congestion and the occurrence of secondary incidents. A fuzzy entropy feature selection methodology is applied first in order to determine redundant factors and rank factor importance with respect to their contribution on the predictability of incident duration. Neural network models for incident duration prediction with single and competing uncertainties are then developed. The results presented in the article indicate that alignment, collision type, and downstream geometry may be considered as redundant when modeling incident duration. The article discusses how rainfall intensity is a highly contributing feature, while lane volume, number of blocked lanes, as well as number of vehicles involved in the incident are among the top ranking factors for determining the extent of duration. The last section of the article shows how joint consideration of severe congestion and secondary incident occurrence may improve the generalization power of the prediction models.

Enhancing Predictions in Signalized Arterials with Information on Short-Term Traffic Flow Dynamics

Short-term traffic flow predictions are an essential part of intelligent transportation systems. Previous research underlines the difficulty in systematically assessing the predictability of traffic flow near capacity or during congested conditions. In this article a neural network prediction scheme is proposed that is consistent with the pattern-based evolution of traffic flow and has the capability of exploiting past information to acquire knowledge on the traffic dynamics in order to enhance predictability. Findings indicate that pattern-based predictions are more accurate—in the traffic flow regimes considered—when compared to other local and global prediction techniques that operate under the time-series consideration. The pattern-based prediction scheme was also found to outperform the other methods tested in the knowledge of the anticipated traffic flow state in all traffic flow conditions considered.

Pattern-Based Short-Term Urban Traffic Predictor

Short-term urban traffic predictor is a prediction system based on artificial intelligence and advanced analysis of nonlinear dynamics of short-term traffic flow. The aim is to address: (1) the need for accurate anticipated traffic information, (2) the ability to cope with various traffic conditions in an integrated up-to-date functional structure and (3) the need for accurate and timely predictive short-term traffic information in an extended time horizon in cases of data collection malfunction. The conceptual and functional framework of short-term urban traffic predictor is presented. The above are encapsulated in a user-friendly application. Finally, some interesting results regarding computational time and accuracy of recursive predictions are presented

A Real-Time Parking Prediction System for Smart Cities

A methodological framework for multiple steps ahead parking availability prediction is presented. Two different types of predictions are provided: the probability of a free space to continue being free in subsequent time intervals, and the short-term parking occupancy prediction in selected regions of an urban road network. The available data come from a wide network of on-street parking sensors in the “smart” city of Santander, Spain. The sensor network is segmented in four different regions, and then survival and neural network models are developed for each region separately. Findings show that the Weibull parametric models best describe the probability of a parking space to continue to be free in the forthcoming time intervals. Moreover, simple genetically optimized multilayer perceptrons accurately predict region parking occupancy rates up to 30 minutes in the future by exploiting 1-minute data. Finally, the real time, Web-based, implementation of the proposed parking prediction availability system is presented.

Testing and Comparing Neural Network and Statistical Approaches for Predicting Transportation Time Series

Univariate and multivariate neural network (NN) and autoregressive time series models are compared with regard to application to the short-term forecasting of freeway speeds. Statistical tests are used to evaluate the developed models with respect to temporal data resolution, prediction accuracy, and quality of fit. The results indicate that, by and large, NNs provide more accurate predictions than do classical statistical approaches, particularly for finer data resolutions. Evaluation of model fit indicated that, in contrast to vector autoregressive models, NNs may also provide unbiased predictions. Overall, the findings clearly suggest the need to jointly consider statistical and NN models to develop more efficient prediction models.