Towards a unified understanding of uncertainty quantification in traffic flow forecasting

Qian, W., Zhao, Y., Zhang, D., Chen, B. , Zheng, K. and Zhou, X. (2024) Towards a unified understanding of uncertainty quantification in traffic flow forecasting. IEEE Transactions on Knowledge and Data Engineering, 36(5), pp. 2239-2256. (doi: 10.1109/TKDE.2023.3312261)

Abstract

Uncertainty is an essential consideration for time series forecasting tasks. In this work, we focus on quantifying the uncertainty of traffic forecasting from a unified perspective. We develop a novel traffic forecasting framework, namely Deep Spatio-Temporal Uncertainty Quantification (DeepSTUQ), which can estimate both aleatoric and epistemic uncertainty. Specifically, we first leverage a spatio-temporal model to model the complex spatio-temporal correlations of traffic data. Subsequently, two independent sub-neural networks maximizing the heterogeneous log-likelihood are developed to estimate aleatoric uncertainty. To estimate epistemic uncertainty, we combine the merits of variational inference and deep ensembling by integrating the Monte Carlo dropout and the Adaptive Weight Averaging re-training methods, respectively. Furthermore, to relax the Gaussianity assumption, mitigate overfitting, and improve horizon-wise uncertainty quantification performance, we define a new calibration method called Multi-horizon Conformal Calibration (MHCC). Finally, we provide a theoretical analysis of the proposed unified approach based on the PAC-Bayes theory. Extensive experiments are conducted on four public datasets, and the empirical results suggest that the proposed method outperforms state-of-the-art methods in terms of both point prediction and uncertainty quantification.

• Item Type: Articles
• Additional Information: This work is partially supported by NSFC (No. 61972069, 61832017, 62272086), Shenzhen Municipal Science and Technology R&D Funding Basic Research Program (JCYJ20210324133607021), Municipal Government of Quzhou under Grant No. 2022D037, and Key Laboratory of Data Intelligence and Cognitive Computing, Longhua District, Shenzhen.
• Status: Published
• Refereed: Yes
• Glasgow Author(s) Enlighten ID: Chen, Dr Bowei
• Authors: Qian, W., Zhao, Y., Zhang, D., Chen, B., Zheng, K., and Zhou, X.
• College/School: College of Social Sciences > Adam Smith Business School > Management
• Journal Name: IEEE Transactions on Knowledge and Data Engineering
• Publisher: IEEE
• ISSN: 1041-4347
• ISSN (Online): 1558-2191
• Published Online: 06 September 2023
• Copyright Holders: Copyright © 2023 IEEE
• First Published: First published in IEEE Transactions on Knowledge and Data Engineering 36(5):2239-2256
• Publisher Policy: Reproduced in accordance with the copyright policy of the publisher