HOMGNN

This is a PyTorch implementation of the paper: Higher-Order Masked Graph Neural Networks for Multi-Step Traffic Flow Prediction.

Requirements

The model is implemented using Python3 with dependencies specified in requirements.txt

Data Preparation

Traffic datasets

Download the METR-LA and PEMS-BAY dataset from Google Drive or Baidu Yun provided by Li et al. . Move them into the data folder.

# Create data directories
mkdir -p data/{METR-LA,PEMS-BAY}

# METR-LA
python generate_training_data.py --output_dir=data/METR-LA --traffic_df_filename=data/metr-la.h5

# PEMS-BAY
python generate_training_data.py --output_dir=data/PEMS-BAY --traffic_df_filename=data/pems-bay.h5

Model Training

Retrain Model on:

• METR-LA

python main.py --adj_data ./data/sensor_graph/adj_mx.pkl --data ./data/METR-LA --num_nodes 207 --order 2  --neiaccount 2

• PEMS-BAY

python main.py --adj_data ./data/sensor_graph/adj_mx_bay.pkl --data ./data/PEMS-BAY --num_nodes 325 --order 2 --neiaccount 1

Citation

@inproceedings{yuan2022higher,
  title={Higher-Order Masked Graph Neural Networks for Multi-Step Traffic Flow Prediction},
  author={Kaixin Yuan, Jing Liu, and Jian Lou},
  booktitle={22nd IEEE International Conference on Data Mining},
  year={2022}
}

Our code is based on the implementation of MTGNN .