Hybrid Learning Model of Global–Local Graph Attention Network and XGBoost for Inferring Origin–Destination Flows DOI Creative Commons

Zhenyu Shan,

Fei Yang, Xian Shi

et al.

ISPRS International Journal of Geo-Information, Journal Year: 2025, Volume and Issue: 14(5), P. 182 - 182

Published: April 24, 2025

Origin–destination (OD) flows are essential for urban studies, yet their acquisition is often hampered by high costs and privacy constraints. Prevailing inference methodologies inadequately address latent spatial dependencies between non-contiguous distant areas, which useful understanding modern transportation systems with expanding regional interactions. To these challenges, this paper propose a hybrid learning model the Global–Local Graph Attention Network XGBoost (GLGAT-XG) to infer OD from both global local geographic contextual information. First, we represent study area as an undirected weighted graph. Second, design GLGAT encode correlation feature information into embeddings within multitask setup. Specifically, employs graph transformer capture correlations attention network extract followed fusion ensure validity. Finally, flow performed based on GLGAT-generated embeddings. The experimental results of multiple real-world datasets demonstrate 8% improvement in RMSE, 7% MAE, 10% CPC over baselines. Additionally, produce multi-scale dataset Xian, China, further reveal spatial-scale effects. This research builds existing offers significant practical implications planning sustainable development.

Language: Английский

A novel approach for cluster detection in trajectory data with low cluster-to-noise density ratio DOI
Zidong Fang, Tao Pei, Xiaohan Liu

et al.

International Journal of Geographical Information Science, Journal Year: 2025, Volume and Issue: unknown, P. 1 - 30

Published: April 30, 2025

Language: Английский

Citations

0
Hybrid Learning Model of Global–Local Graph Attention Network and XGBoost for Inferring Origin–Destination Flows DOI Creative Commons

Zhenyu Shan,

Fei Yang, Xian Shi

et al.

ISPRS International Journal of Geo-Information, Journal Year: 2025, Volume and Issue: 14(5), P. 182 - 182

Published: April 24, 2025

Origin–destination (OD) flows are essential for urban studies, yet their acquisition is often hampered by high costs and privacy constraints. Prevailing inference methodologies inadequately address latent spatial dependencies between non-contiguous distant areas, which useful understanding modern transportation systems with expanding regional interactions. To these challenges, this paper propose a hybrid learning model the Global–Local Graph Attention Network XGBoost (GLGAT-XG) to infer OD from both global local geographic contextual information. First, we represent study area as an undirected weighted graph. Second, design GLGAT encode correlation feature information into embeddings within multitask setup. Specifically, employs graph transformer capture correlations attention network extract followed fusion ensure validity. Finally, flow performed based on GLGAT-generated embeddings. The experimental results of multiple real-world datasets demonstrate 8% improvement in RMSE, 7% MAE, 10% CPC over baselines. Additionally, produce multi-scale dataset Xian, China, further reveal spatial-scale effects. This research builds existing offers significant practical implications planning sustainable development.

Language: Английский

Citations

0