ST-GAT Resident OD Prediction Model Based on Mobile Signaling Data

Aiming at the challenges of spatio-temporal heterogeneity and dynamic correlation in the prediction of travel origins and destinations for urban residents, this study invokes a spatio-temporal graph attention network (ST-GAT) OD prediction model based on cell phone signaling data. This model innovat...

Full description

Saved in:
Bibliographic Details
Main Authors: Xianguang Jia, Weijie Fang, Yingying Lyu, Jinke Zhang, Mengyi Guo, Dong Li, Jie Qu, Fengxiang Guo
Format: Article
Language:English
Published: IEEE 2025-01-01
Series:IEEE Access
Subjects:
Online Access:https://ieeexplore.ieee.org/document/11028587/
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Aiming at the challenges of spatio-temporal heterogeneity and dynamic correlation in the prediction of travel origins and destinations for urban residents, this study invokes a spatio-temporal graph attention network (ST-GAT) OD prediction model based on cell phone signaling data. This model innovatively introduces the graph attention mechanism into the spatio-temporal graph network (ST-GNN), in the spatial dimension, the attention layer (GAL) dynamically learns the attention weights among nodes to adaptively capture the dynamic spatial dependencies in the transportation network, and in the temporal dimension, the temporal convolutional layer extracts the multiscale temporal patterns, which efficiently captures the complex spatiotemporal dependencies in the OD data. The data cleaning innovatively uses the DTW data cleaning model, and MSE, MAE, RMSE are used as evaluation indexes. In this paper, this model is compared with the existing ST-GCN, DMS, GMM, PSAM-CNN, ST-Transformer, and COMD models, and the experimental results show that the predicted values of the ST-GAT model have a significant improvement in the prediction accuracy compared to other models. The research in this paper provides an effective solution for the application of Intelligent Transportation System (ITS) and demonstrates the potential of Deep Learning (DL) in urban transportation planning.
ISSN:2169-3536