STPGNN
June 4, 2024 · View on GitHub
This is an implementation of [Spatio-Temporal Pivotal Graph Neural Networks for Traffic Flow Forecasting]
Environment
- python 3.7.4
- torch 1.2.0
- numpy 1.17.2
Dataset
Step 1: Download the processed dataset from Baidu Yun (Access Code:luck).
Step 2: Put them into data directories.
Train command
# Train with PEMS03
python train.py --data PEMS03 --topk 70
# Train with PEMS04
python train.py --data PEMS04 --topk 70 --dims 64
# Train with PEMS07 (Reduce the batch_size appropriately if don't have enough memory.)
python train.py --data PEMS07 --topk 175
# Train with PEMS08
python train.py --data PEMS08 --topk 35
# Train with ENG
python train.py --data ENG --topk 35
# Train with TaxiBJ
python train.py --data bjtaxi --topk 200