New method improves forecasting of traffic and water quality patterns
Decompose Dynamics Before Learning Dependencies in Spatiotemporal Systems
Computational Engineering, Finance, and Science
Summary
Many systems, like city traffic or river water quality, change over time and space in complicated ways. The authors created CANDOR, a way to first split the system’s changes into parts before figuring out how different parts influence each other. This helps to better understand and predict how local events happen and spread through the network. Tests on traffic and water quality data showed this method predicts future states more accurately than other methods.
What this means in practice
- •For urban traffic planners: Improve predictions of traffic flow and congestion by modeling how local traffic changes and propagates through networks more accurately.
- •For environmental monitoring teams: Enhance forecasting of water quality changes in rivers by separating local effects and their spread in complex natural systems.
Authors
Ziqi Wang, Daojiang Hu, Cheng Bao, Zhiwei Ling, Wenzhuo Qian, Jiahui Zhai, Hailiang Zhao
Abstract
Relations in networked spatiotemporal systems are often learned from observations that entangle dynamics governed by different mechanisms, obscuring what evolves locally and how it propagates across nodes. We introduce Component-Aware Network Dynamics with Ordered Relations (CANDOR), which decomposes local dynamics before learning their dependencies. CANDOR represents each trajectory through a persistent background, gradual accumulation and release, and sparse shocks. Conditioned on these components, a delay-aware physical branch models edge and sample-dependent propagation over directed topology, while a topology-unconstrained functional branch discovers latent dependencies from background dynamics. Context-adaptive fusion combines functional, forward-propagation, and reverse-support forecasts, with training objectives encouraging specialized and semantically consistent representations. Experiments on two traffic benchmarks and three long-horizon water-quality datasets span two distinct spatiotemporal systems: human-driven urban traffic and naturally evolving river water quality. CANDOR consistently outperforms the strongest baselines, reducing MAE by up to 4.31% for traffic and MSE by up to 5.47% for water-quality forecasting. These results establish decomposition before dependency learning as an effective principle for spatiotemporal representation learning.