This is an official implementation of WFTNet: Exploiting Global and Local Periodicity in Long-term Time Series Forecasting.

Ensure you are using Python 3.9 and install the necessary dependencies by running:

pip install -r requirements.txt

Download data from AutoFormer. Put all data into a seperate folder ./dataset and make sure it has the following structure:

dataset
├── electricity
│   └── electricity.csv
├── ETT-small
│   ├── ETTh1.csv
│   ├── ETTh2.csv
│   ├── ETTm1.csv
│   └── ETTm2.csv
├── traffic
│   └── traffic.csv
└── weather
    └── weather.csv

The training scripts for all datasets are located in the ./scripts directory.

To train a model using the ETTh1 dataset:

1. Navigate to the repository's root directory.
2. Execute the following command:

   sh ./scripts/ETTh1.sh

Upon completion of the training:

• The trained model will be saved in the ./checkpoints directory.
• Visualization outputs can be found in ./test_results.
• Numerical results in .npy format are located in ./results.
• A summary of the quantitative metrics is available in ./results.txt.

If you find this repo useful, please cite our paper as follows:

@article{liu2023wftnet,
  title={WFTNet: Exploiting Global and Local Periodicity in Long-term Time Series Forecasting},
  author={Liu, Peiyuan and Wu, Beiliang and Li, Naiqi and Dai, Tao and Lei, Fengmao and Bao, Jigang and Jiang, Yong and Xia, Shu-Tao},
  journal={arXiv preprint arXiv:2309.11319},
  year={2023}
}

If you have any questions, please contact [email protected] or submit an issue.

We appreciate the following repo for their code and dataset:

• https://github.com/thuml/Time-Series-Library
• https://github.com/thuml/Autoformer