Implementation of WaveBound: Dynamic Error Bounds for Stable Time Series Forecasting (NeurIPS 2022), applied to Graph WaveNet for Deep Spatial-Temporal Graph Modeling, IJCAI 2019. This repo is officially provided as examplary code for applying WaveBound method to Traffic domain. To view WaveBound's official implementation, visit the wavebound-github-repo.
# 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
sh scripts/metr_la.sh # METR-LA
sh scripts/pems_bay.sh # PEMS-BAY
STEPS |
15min |
|
|
30min |
|
|
60min |
|
|
Metrics |
MAE |
RMSE |
MAPE |
MAE |
RMSE |
MAPE |
MAE |
RMSE |
MAPE |
GWNet |
2.6998 |
5.1684 |
0.0693 |
3.0966 |
6.2413 |
0.0837 |
3.5812 |
7.4361 |
0.1010 |
GWNet+WaveBound |
2.6703 |
5.1104 |
0.0691 |
3.0313 |
6.1046 |
0.0832 |
3.4478 |
7.1474 |
0.0990 |
STEPS |
15min |
|
|
30min |
|
|
60min |
|
|
Metrics |
MAE |
RMSE |
MAPE |
MAE |
RMSE |
MAPE |
MAE |
RMSE |
MAPE |
GWNet |
1.3003 |
2.7225 |
0.0273 |
1.6171 |
3.6546 |
0.0364 |
1.9201 |
4.4375 |
0.0449 |
GWNet+WaveBound |
1.2917 |
2.7221 |
0.0269 |
1.6080 |
3.6538 |
0.0359 |
1.9045 |
4.4103 |
0.0449 |