GeoFlow: Geo-Aware Modeling of Inter-Area Relationships in Origin-Destination Flow Prediction and Generation

2026-07-06

AI Summary

This work addresses the limitation of existing graph-based models in origin–destination (OD) flow prediction, which often neglect geographical attributes and struggle to capture long-range and cross-regional dependencies. To overcome this, the authors propose a novel architecture that integrates geographical priors by employing a coordinate-aware encoder to embed spatial features—such as relative positions, k-hop distances, and geodesic distances—into geometrically informed and intrinsically rich regional representations. Furthermore, they introduce an axial-global attention decoder that, for the first time, adapts flow-matching models to the OD flow generation task. Experimental results demonstrate that the proposed method significantly outperforms state-of-the-art approaches in terms of prediction accuracy, generation fidelity, and diversity, while ablation studies confirm the effectiveness of each architectural component.

Abstract

Origin-destination (OD) flow modeling underpins urban planning and mobility analysis, but prevailing graph-based methods often neglect salient geographic attributes, limiting their ability to model long-range and multi-area dependencies. In this paper, we introduce GeoFlow, a novel framework that (i) augments area representations with geospatial attributes, including relative positions, k-hop and geodesic distances, (ii) employs a specialized geometric-intrinsic fusion encoder design that combines graph attention for intrinsic area signals with coordinate-aware encoders for global structure, and (iii) adopts an axial-global attention decoder to capture OD-specific competitive dependencies. For OD flow generation, GeoFlow is paired with flow matching models to produce more authentic and diverse mobility samples. Empirically, GeoFlow achieves superior performance in predictive accuracy, while substantially improving generative fidelity and diversity. Ablation and analytical studies confirm the contribution of each component. Code is available at https://github.com/ZheruiHuang/GeoFlow.

Problem

• origin-destination flow
• geospatial attributes
• long-range dependencies
• multi-area relationships
• graph-based methods

Innovation

• geo-aware modeling
• origin-destination flow prediction
• geometric-intrinsic fusion
• axial-global attention
• flow generation