Graph convolutional networks for traffic forecasting with missing values

• Published in: Data mining and knowledge discovery Vol. 37; no. 2; pp. 913 - 947
• Main Authors: Zuo, Jingwei, Zeitouni, Karine, Taher, Yehia, Garcia-Rodriguez, Sandra
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.03.2023; Springer Nature B.V; Springer
• Subjects: Artificial Intelligence; Artificial neural networks; Chemistry and Earth Sciences; Computer Science; Condensed Matter; Context; Data Mining and Knowledge Discovery; Datasets; Deep learning; Disordered Systems and Neural Networks; Experiments; Forecasting; Graph neural networks; Graph representations; Information Storage and Retrieval; Mathematical models; Neural networks; Physics; Sensors; Special Issue of the Journal Track of ECML PKDD 2022; Statistics for Engineering; Task complexity; Traffic information; Deep learning; Memory networks; Graph convolutional networks; Traffic forecasting; Neural networks; Missing values; Artificial intelligence
• ISSN: 1384-5810; 1573-756X
• DOI: 10.1007/s10618-022-00903-7

Traffic forecasting has attracted widespread attention recently. In reality, traffic data usually contains missing values due to sensor or communication errors. The Spatio-temporal feature in traffic data brings more challenges for processing such missing values, for which the classic techniques (e.g., data imputations) are limited: (1) in temporal axis, the values can be randomly or consecutively missing; (2) in spatial axis, the missing values can happen on one single sensor or on multiple sensors simultaneously. Recent models powered by Graph Neural Networks achieved satisfying performance on traffic forecasting tasks. However, few of them are applicable to such a complex missing-value context. To this end, we propose GCN-M, a Graph Convolutional Network model with the ability to handle the complex missing values in the Spatio-temporal context. Particularly, we jointly model the missing value processing and traffic forecasting tasks, considering both local Spatio-temporal features and global historical patterns in an attention-based memory network. We propose as well a dynamic graph learning module based on the learned local-global features. The experimental results on real-life datasets show the reliability of our proposed method.