APSIPA Transactions on Signal and Information Processing > Vol 13 > Issue 1

POI-based Double-deck Graph Convolution Network for Traffic Forecasting

Xinhao Zhao, University of Chinese Academy of Sciences, China, Zhe Wu, Peng Cheng Laboratory, China, wuzh02@pcl.ac.cn , Xinfeng Zhang, University of Chinese Academy of Sciences, China, Zhiyuan Deng, University of Chinese Academy of Sciences, China, Li Su, University of Chinese Academy of Sciences, Peng Cheng Laboratory, and Chinese Academy of Sciences, China, suli@ucas.ac.cn , Guorong Li, University of Chinese Academy of Sciences and Chinese Academy of Sciences, China, Yaowei Wang, Peng Cheng Laboratory, China, Qingming Huang, Chinese Academy of Sciences and University of Chinese Academy of Sciences, China
 
Suggested Citation
Xinhao Zhao, Zhe Wu, Xinfeng Zhang, Zhiyuan Deng, Li Su, Guorong Li, Yaowei Wang and Qingming Huang (2024), "POI-based Double-deck Graph Convolution Network for Traffic Forecasting", APSIPA Transactions on Signal and Information Processing: Vol. 13: No. 1, e10. http://dx.doi.org/10.1561/116.20230081

Publication Date: 04 Jul 2024
© 2024 X. Zhao, Z. Wu, X. Zhang, Z. Deng, L. Su, G. Li, Y. Wang and Q. Huang
 
Subjects
Classification and prediction,  Data mining,  Graphical models,  Multimodal signal processing
 
Keywords
Traffic forecastingdynamic graph constructionspatial attention mechanism
 

Share

Open Access

This is published under the terms of CC BY-NC.

Downloaded: 242 times

In this article:
Introduction 
Related Work 
Preliminaries 
Methodology 
Experiments 
Conclusions 
References 

Abstract

Traffic flow forecasting, a vital task of multivariate time series prediction, has recently expanded to incorporate Points of Interest (POI) as an additional source of data. Rather than merely leveraging historical traffic flows, POIs facilitate the understanding of inherent geographical connections and potential functional interactions between nodes. However, traditional POI-based methods tend to use POI as static feature embeddings to compute functional similarity matrices, failing to consider the dynamic influence of node functionality on traffic patterns. This overlooks the reality that even regions with analogous POIs can exhibit fluctuating traffic flow trends, particularly over extended periods. In this paper, we propose the POI-based Double-deck Graph Convolution Network (PDGCN) for more nuanced traffic forecasting. To identify potential POI-based traffic patterns, we employ the spectral clustering method to group nodes with comparable POI functionalities into regions. We then devise a POI-based dynamic graph module with temporal convolution and attention mechanisms to trace the evolving relationships between traffic nodes. This novel design underpins regional features. Experiments on two real datasets demonstrate that PDGCN effectively detects dynamic functional relationships between nodes and delivers superior prediction accuracy.

DOI:10.1561/116.20230081