GeoDI-Lab / HyGCN

HyGCN

Title:

A Hypergraph-based Hybrid Graph Convolutional Network for Intracity Human Activity Intensity Prediction and Geographic Relationship Interpretation

Introduction:

This is a Pytorch implementation of HyGCN. Our code is based on ASTGCN (https://github.com/guoshnBJTU/ASTGCN-r-pytorch) and Pytorch Geometric (https://github.com/pyg-team/pytorch_geometric).

Pre:

Step1: Clone the code of ASTGCN.

Step2: Put HyGCN.py into model and MN_astgcn.yaml into configurations.

Datasets:

Step1: Download the demo dataset (MN_demo.npz). And Put it into folder data (If not, please create it).

Step2: Process dataset like ASTGCN.

python prepareData.py --config configurations/MN_astgcn.conf

Train and Test:

Please refer to ASTGCN's Run and Test (https://github.com/guoshnBJTU/ASTGCN-r-pytorch).

Reference:

Wang, Yi and Zhu, Di*. "A Hypergraph-based Hybrid Graph Convolutional Network for Intracity Human Activity Intensity Prediction and Geographic Relationship Interpretation." Information Fusion (2023): 102149. https://doi.org/10.1016/j.inffus.2023.102149