High‐performance traffic speed forecasting based on spatiotemporal clustering of road segments

Abstract

Traffic speed prediction is an indispensable element of intelligent transportation systems. Numerous studies have devoted to high-precision prediction models. However, most existing methods implement the link-wise or network-wide input. The former is timeconsuming especially for large-scale applications, while the latter may incur the dilemma of underfitting owing to the heterogeneous traffic states within the entire network. Herein, we propose a novel prediction scheme based on spatiotemporal traffic pattern clustering. Firstly, road segments are partitioned into several groups via the developed clustering approach, which considers both the observed data sequence and spatial topology structure. Subsequently, sequence-to-sequence learning architecture is employed for each group to generate predictions for the entire traffic network. Validated by a real-world dataset in Beijing, our proposed paradigm offers a significant improvement over other well-known benchmarks for various prediction intervals in terms of prediction accuracy and computational efficiency. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.

A comprehensive study of speed prediction in transportation system: From vehicle to traffic

“…Meanwhile, Bayesian optimization is regarded as an effective approach for parameter optimization in ( Zhang et al., 2020b ). In addition, the Encoder-Decoder structure, usually consisting of the RNNs, is employed to handle the issue of different lengths of input and output sequences ( Li et al., 2018b ; Zhang et al., 2019 , 2020d ). However, the limitation of Encoder-Decoder is obvious.…”

Spatiotemporal traffic data imputation by synergizing low tensor ring rank and nonlocal subspace regularization

“…where G ≠n [2] ∈ ℝ ∏N j =1, j ≠n I j ×R n R n+1 denotes the reversed mode-2 unfolding of  ≠n ∈ ℝ R n+1 × ∏N j =1, j ≠n I j ×R n (the multilinear product of all factors except the factor  n ).…”

“…Spatiotemporal traffic data, gathered from various data collection devices (e.g. mobile sensors and GPS devices), play a key role in traffic management and urban planning [1][2][3][4][5][6][7]. In practice, due to the transmission distortion and natural weather, traffic data is usually incomplete, resulting in seriously affecting its subsequent applications.…”

Lane‐level short‐term travel speed prediction for urban expressways: An attentive spatio‐temporal deep learning approach

“…the gated recurrent unit (GRU)) [24,25]. The well-known RNN-based structure sequence-to-sequence network has also been proposed by Zhang et al to consider the link-wise or network-wise travel speed prediction [26]. Based on representing the correlations of traffic flows with two-dimensional matrices as inputs, CNN-based models have become a new trend in traffic flow forecasting [17].…”