Evaluation of Recent Computational Approaches in Short-Term Traffic Forecasting

Abstract

Part 3: Computational Intelligence and AlgorithmsInternational audienceComputational technologies under the domain of intelligent systems are expected to help the rapidly increasing traffic congestion problem in recent traffic management. Traffic management requires efficient and accurate forecasting models to assist real time traffic control systems. Researchers have proposed various computational approaches, especially in short-term traffic flow forecasting, in order to establish reliable traffic patterns models and generate timely prediction results. Forecasting models should have high accuracy and low computational time to be applied in intelligent traffic management. Therefore, this paper aims to evaluate recent computational modeling approaches utilized in short-term traffic flow forecasting. These approaches are evaluated by real-world data collected on the British freeway (M6) from 1st to 30th November in 2014. The results indicate that neural network model outperforms generalized additive model and autoregressive integrated moving average model on the accuracy of freeway traffic forecasting

A Taxonomy of Traffic Forecasting Regression Problems From a Supervised Learning Perspective

“…It does generate, however, issues related to the selection of the proper data-driven method. The overall experience in multi-target modelling points out the use of non-parametric techniques, such as NNs [56], [115], to predict fundamental macroscopic traffic variables together with travel time [5], [102], [128].…”

“…The majority of the transportation literature categorized has approached TF problems that contain exclusively this type of data source. Specifically, only three papers [19], [119], [128] highlighted in Tables 1 and 2 include more than one kind of data source. In this sense, the integration of multiple data sources is an opportunity from the transportation perspective and a challenge in the ML area.…”

“…From the perspective of how many traffic parameters are predicted, the taxonomy demonstrates that most TF problems are focused on single-target output predictions (stacked bar 4.2 Output Modelling). The latter opens the possibility to explore the benefits and challenges of multiple output targets, which according to Tables 1 and 2 has been approached by very few studies [5], [102], [128].…”

Joint Feature Selection and Parameter Tuning for Short-Term Traffic Flow Forecasting Based on Heuristically Optimized Multi-layer Neural Networks

“…In this context, ANNs and their combination with other methods have been extensively used with relative success over naïve (historic average and last measurement predictions) and time-series models [3][4][5][6][7], fueled by prior literature evincing that ANNs are more responsive to changes in data [8]. However, neural networks behave in a black-box manner that hinders their understanding.…”