PAST: A Primary-Auxiliary Spatio-Temporal Network for Traffic Time Series Imputation

2025-11-17

AI Summary

Traffic time series imputation faces challenges from diverse missing patterns—including random, fiber-wise, and block-wise gaps—where existing methods struggle to jointly achieve missing-location adaptivity and effective long-range spatiotemporal dependency modeling. To address this, we propose PAST, the first framework that disentangles traffic dynamics into a *principal pattern* (intrinsic spatiotemporal correlations) and an *auxiliary pattern* (externally driven factors), implemented via a dual-module collaborative architecture: the Graph Integration Module (GIM) employs interval-aware dropout and multi-order convolutions to model the principal pattern; the Cross-Gated Module (CGM) applies bidirectional gating over timestamps, node attributes, and other exogenous features to extract the auxiliary pattern. An ensemble self-supervised learning strategy is further introduced to enhance robustness. Extensive experiments across three real-world datasets and 27 missing scenarios demonstrate that PAST outperforms seven state-of-the-art baselines, achieving up to 26.2% reduction in RMSE and 31.6% reduction in MAE.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Machine Learning: Time-Series/Data Streams
• Computer Vision: Diffusion Models for Vision

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

Traffic time series imputation is crucial for the safety and reliability of intelligent transportation systems, while diverse types of missing data, including random, fiber, and block missing make the imputation task challenging. Existing models often focus on disentangling and separately modeling spatial and temporal patterns based on relationships between data points. However, these approaches struggle to adapt to the random missing positions, and fail to learn long-term and large-scale dependencies, which are essential in extensive missing conditions. In this paper, patterns are categorized into two types to handle various missing data conditions: primary patterns, which originate from internal relationships between data points, and auxiliary patterns, influenced by external factors like timestamps and node attributes. Accordingly, we propose the Primary-Auxiliary Spatio-Temporal network (PAST). It comprises a graph-integrated module (GIM) and a cross-gated module (CGM). GIM captures primary patterns via dynamic graphs with interval-aware dropout and multi-order convolutions, and CGM extracts auxiliary patterns through bidirectional gating on embedded external features. The two modules interact via shared hidden vectors and are trained under an ensemble self-supervised framework. Experiments on three datasets under 27 missing data conditions demonstrate that the imputation accuracy of PAST outperforms seven state-of-the-art baselines by up to 26.2% in RMSE and 31.6% in MAE.

Problem

• Handling diverse missing data types in traffic time series imputation
• Learning long-term dependencies under extensive missing data conditions
• Adapting to random missing positions in spatio-temporal data

Innovation

• Primary-Auxiliary Spatio-Temporal network handles diverse missing data
• Graph-integrated module captures primary patterns via dynamic graphs
• Cross-gated module extracts auxiliary patterns through bidirectional gating