C3DE: Causal-Aware Collaborative Neural Controlled Differential Equation for Long-Term Urban Crowd Flow Prediction

2025-09-15

AI Summary

Long-term urban crowd flow forecasting faces three key challenges: (1) accumulating sampling errors degrade prediction accuracy over extended horizons; (2) points-of-interest (POIs) evolve dynamically and exhibit asynchronous, multi-timescale dependencies with crowd flows; and (3) spurious correlations between POIs and flows hinder causal inference. To address these, we propose C3DE—a causal-aware, end-to-end forecasting framework. C3DE employs a dual-path Neural Controlled Differential Equation (NCDE) to jointly model multi-scale asynchronous dynamics; integrates a counterfactual causal effect estimator to dynamically identify and correct for true POI causal influences while suppressing spurious correlation accumulation; and fuses heterogeneous spatiotemporal data sources. Evaluated on three real-world urban datasets, C3DE consistently outperforms state-of-the-art methods, demonstrating superior accuracy and robustness—especially in regions with highly volatile crowd flows.

Technology Category

• Reasoning under Uncertainty: Causality
• Machine Learning: Causal Learning
• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Long-term urban crowd flow prediction suffers significantly from cumulative sampling errors, due to increased sequence lengths and sampling intervals, which inspired us to leverage Neural Controlled Differential Equations (NCDEs) to mitigate this issue. However, regarding the crucial influence of Points of Interest (POIs) evolution on long-term crowd flow, the multi-timescale asynchronous dynamics between crowd flow and POI distribution, coupled with latent spurious causality, poses challenges to applying NCDEs for long-term urban crowd flow prediction. To this end, we propose Causal-aware Collaborative neural CDE (C3DE) to model the long-term dynamic of crowd flow. Specifically, we introduce a dual-path NCDE as the backbone to effectively capture the asynchronous evolution of collaborative signals across multiple time scales. Then, we design a dynamic correction mechanism with the counterfactual-based causal effect estimator to quantify the causal impact of POIs on crowd flow and minimize the accumulation of spurious correlations. Finally, we leverage a predictor for long-term prediction with the fused collaborative signals of POI and crowd flow. Extensive experiments on three real-world datasets demonstrate the superior performance of C3DE, particularly in cities with notable flow fluctuations.

Problem

• Mitigating cumulative sampling errors in long-term urban crowd flow prediction
• Modeling multi-timescale asynchronous dynamics between crowd flow and POI evolution
• Quantifying causal impact of POIs while minimizing spurious correlation accumulation

Innovation

• Dual-path NCDE backbone captures multi-timescale evolution
• Counterfactual causal estimator quantifies POI impact on flow
• Fused collaborative signals enable long-term crowd prediction