🔥 News

• 2025/05/27: Code for TimePro is available.

📷 Introduction

1️⃣ A powerful and efficient multivariate time series forecasting model

2️⃣ Architecture

3️⃣ Performance

📚 Use example

• Environment

  conda create --name timepro python=3.9 -y
  conda activate timepro
  pip install -r requirements.txt
  cd selective_scan && pip install .
  

• Dataset

  The dataset can be download at this link

  ln -s /path/to/TimePro_dataset/ dataset
  

• Train

  bash scripts/TimePro_ECL.sh
  bash scripts/TimePro_Exchange.sh
  bash scripts/TimePro_SolarEnergy.sh
  bash scripts/TimePro_Weather.sh
  bash scripts/TimePro_ETTh1.sh
  bash scripts/TimePro_ETTh2.sh
  bash scripts/TimePro_ETTm1.sh
  bash scripts/TimePro_ETTm2.sh
  

🌟 Citation

If you are interested in our work, please consider giving a 🌟 and citing our work below.

@inproceedings{
timepro,
title={TimePro: Efficient Multivariate Long-term Time Series Forecasting with Variable- and Time-Aware Hyper-state},
author={Xiaowen Ma and Zhen-Liang Ni and Shuai Xiao and Xinghao Chen},
booktitle={Forty-second International Conference on Machine Learning},
year={2025},
url={https://openreview.net/forum?id=s69Ei2VrIW}
}

💡Acknowledgment

Thanks to previous open-sourced repo:

S-Mamba, VMamba, Spatial-Mamba