S2GNN

S2GNN: Long-term Spatio-Temporal Forecasting using Spectral Graph Neural Networks

Dependencies

# Install Python
conda create -n S2GNN python=3.11
conda activate S2GNN
# Install PyTorch
pip install torch==2.3.1 torchvision==0.18.1 torchaudio==2.3.1 --index-url https://download.pytorch.org/whl/cu121
# Install other dependencies
pip install -r requirements.txt

Datasets

Processed datasets and raw data can be downloaded at: https://mega.nz/folder/AeVknA4C#MuQITYW9YPcaRX6w9uk_Hg, then move it to /datasets folder, for example:

/datasets/PEMS04/data.dat

Implementation

python experiments/train.py -c models/S2GNN/PEMS04.py -g 0
python experiments/train.py -c models/S2GNN/Electricity.py -g 0

To run other baselines, for example:

python experiments/train.py -c baselines/iTransformer/PEMS04.py -g 0