This is the official implementation of paper "WindowMixe:Intra-Window and Inter-Window Modeling for Time Series Forecasting". WindowMixer stems from a simple observation: Time series are recorded in a continuous manner, and the information at any given time point relies on the preceding and succeeding time points for a complete representation.
- Install Python 3.10. For convenience, execute the following command.
pip install -r requirements.txt
-
Prepare Data. You can obtain the well pre-processed datasets from [Google Drive]. Then place the downloaded data in the folder
./dataset. -
Train and evaluate model. We provide the experiment scripts for all benchmarks under the folder
./scripts/. You can reproduce the experiment results as the following examples:
# long-term forecast
bash ./scripts/long_term_forecast/ETT_script/WidowMixer_ETTh1.sh
If you have any questions or suggestions, feel free to contact our maintenance team:
- Quangao Liu (liuquangao@sia.cn)
- Ruiqi Li (liruiqi1@sia.cn)
- Maowei Jiang (jiangmaowei@sia.cn)
Or describe it in Issues.
If you find this repo useful, please cite our paper
@inproceedings{liu2024windowmixer,
title={WindowMixe:Intra-Window and Inter-Window Modeling for Time Series Forecasting},
author={Quangao Liu and Ruiqi Li and Maowei Jiang and Wei Yang and Cheng Liang and Zhuozhang Zou},
year={2024},
}
Our code is based on Time Series Library (TSLib)：https://github.com/thuml/Time-Series-Library