Traffic Speed Prediction: An Attention-Based Method

Abstract

Short-term traffic speed prediction has become one of the most important parts of intelligent transportation systems (ITSs). In recent years, deep learning methods have demonstrated their superiority both in accuracy and efficiency. However, most of them only consider the temporal information, overlooking the spatial or some environmental factors, especially the different correlations between the target road and the surrounding roads. This paper proposes a traffic speed prediction approach based on temporal clustering and hierarchical attention (TCHA) to address the above issues. We apply temporal clustering to the target road to distinguish the traffic environment. Traffic data in each cluster have a similar distribution, which can help improve the prediction accuracy. A hierarchical attention-based mechanism is then used to extract the features at each time step. The encoder measures the importance of spatial features, and the decoder measures the temporal ones. The proposed method is evaluated over the data of a certain area in Hangzhou, and experiments have shown that this method can outperform the state of the art for traffic speed prediction.

An attention‐based deep learning model for traffic flow prediction using spatiotemporal features towards sustainable smart city

“…These schemes do not achieved better prediction results. A speed prediction model was designed using the clustering and attention technique 28 that helps to differentiate the traffic situation and feature extraction.…”

Traffic Information Enrichment: Creating Long-Term Traffic Speed Prediction Ensemble Model for Better Navigation through Waypoints

“…Especially in a short-term perspective, due to the fact that traffic data can be random, varied, and nonlinear, according to the authors of [19], parametric approaches based on linear relationships (like ARIMA) are unsuitable for the analysis of nonlinear traffic data. Prediction accuracy with such models under these conditions is lower than with nonparametric approaches [8]. On the other hand, this disadvantage of parametric models is balanced in a long-term perspective where even these models can reach quality results.…”

“…Radical speed reduction may lead to traffic congestion, which has a negative impact on the environment, health of the inhabitants living in surrounding areas, and traffic fluency, which further impacts road users [2][3][4][5]. Traffic prediction in association with the possibilities of processing new (and often open) data is currently one of the emerging areas of interest among researchers all over the world-see, e.g., [6][7][8][9]. To predict traffic speed, Floating Car Data (FCD) [10] is mainly used, as well as other relevant data sources available in individual states across the world.…”

“…As for the long-term prediction, long-term historical data are mainly used [16], and the prediction duration is in days or months. However, the given time framework can vary slightly depending on the authors and the tasks being solved, see [8,17,18].…”

“…These can be divided into three areas, i.e., data-driven statistical methods, machine learning methods, and deep learning methods [17,19]. Alternatively, they can also be divided into the following areas: Parametric approaches, nonparametric approaches, and deep learningbased approaches [8,18,20,21]. Deep learning-based solutions can be classified among machine learning or nonparametric approaches.…”