Awesome

MSGNet (AAAI2024)

Paper Link:MSGNet: Learning Multi-Scale Inter-Series Correlations for Multivariate Time Series Forecasting

Usage

• Train and evaluate MSGNet

  • You can use the following command:sh ./scripts/ETTh1.sh.

• Train your model

  • Add model file in the folder ./models/your_model.py.
  • Add model in the class Exp_Main.

• Flight dataset

  • You can obtain the dataset from Google Drive. Then please place it in the folder ./dataset.

Model

MSGNet employs several ScaleGraph blocks, each encompassing three pivotal modules: an FFT module for multi-scale data identification, an adaptive graph convolution module for inter-series correlation learning within a time scale, and a multi-head attention module for intra-series correlation learning.

Main Results

Forecast results with 96 review window and prediction length {96, 192, 336, 720}. The best result is represented in bold, followed by underline.

Citation

@article{cai2023msgnet,
  title={MSGNet: Learning Multi-Scale Inter-Series Correlations for Multivariate Time Series Forecasting},
  author={Cai, Wanlin and Liang, Yuxuan and Liu, Xianggen and Feng, Jianshuai and Wu, Yuankai},
  journal={arXiv preprint arXiv:2401.00423},
  year={2023}
}

Acknowledgement

We appreciate the valuable contributions of the following GitHub.

• LTSF-Linear (https://github.com/cure-lab/LTSF-Linear)
• TimesNet (https://github.com/thuml/TimesNet)
• Time-Series-Library (https://github.com/thuml/Time-Series-Library)
• MTGnn (https://github.com/nnzhan/MTGNN)
• Autoformer (https://github.com/thuml/Autoformer)
• Informer (https://github.com/zhouhaoyi/Informer2020)