Traffic Prediction considering Multiple Levels of Spatial-temporal Information: A Multi-scale Graph Wavelet-based Approach

TL;DR

The paper tackles traffic state forecasting on heterogeneous transportation networks by introducing MSGWTCN, which fuses multi-scale graph wavelet spatial blocks with a gated temporal convolution network to model local, intermediate, and global spatial dependencies alongside temporal dynamics. By stacking MS-spatial blocks with a dilated, gated TCN, the approach learns hierarchical spatial-temporal features and uses Chebyshev-based graph wavelet approximations to maintain efficiency. Experiments on Seattle and Manhattan datasets show MSGWTCN consistently outperforms baselines, with best configurations using three scales $s\in\{0.85,3.85,5.85\}$, highlighting the value of multi-scale spatial aggregation. The work also demonstrates interpretability through weight analyses and ablation studies, indicating scale-specific roles in capturing locality and inter-node interactions, and suggests practical guidance for scale selection across network configurations. Overall, the proposed method advances accurate, scalable traffic prediction in complex road networks and provides insights into multi-scale spatial influence on congestion evolution.

Abstract

Although traffic prediction has been receiving considerable attention with a number of successes in the context of intelligent transportation systems, the prediction of traffic states over a complex transportation network that contains different road types has remained a challenge. This study proposes a multi-scale graph wavelet temporal convolution network (MSGWTCN) to predict the traffic states in complex transportation networks. Specifically, a multi-scale spatial block is designed to simultaneously capture the spatial information at different levels, and the gated temporal convolution network is employed to extract the temporal dependencies of the data. The model jointly learns to mount multiple levels of the spatial interactions by stacking graph wavelets with different scales. Two real-world datasets are used in this study to investigate the model performance, including a highway network in Seattle and a dense road network of Manhattan in New York City. Experiment results show that the proposed model outperforms other baseline models. Furthermore, different scales of graph wavelets are found to be effective in extracting local, intermediate and global information at the same time and thus enable the model to learn a complex transportation network topology with various types of road segments. By carefully customizing the scales of wavelets, the model is able to improve the prediction performance and better adapt to different network configurations.

Figures (9)

• Figure 1: Framework of multi-scale graph wavelet-based gated temporal convolution network
• Figure 2: Multiple layers of causal dilation convolution
• Figure 3: Interaction between the primary node and other nodes at (a) small scale, (b) medium scale and (c) high scale. Value of node represents the interaction between the node and its neighboring node, the larger the value is, the higher the interaction between the nodes.
• Figure 4: Network layout of Seattle and New York City.
• Figure 5: Prediction performance (MAE) to different values of scaling parameter