KAN4TSF

KAN4TSF is an official PyTorch implementation of KAN4TSF: Are KAN and KAN-based models Effective for Time Series Forecasting? Although it is called KAN4TSF, it also supports time series forecasting methods of various network structures (CNN, Linear, Transformer and others).

🚩 News (2024.09) Model Zoo: RMoK, NLinear, DLinear, RLinear, PatchTST, iTransformer, STID, TimeLLM

🚩 News (2024.09) Introduction and Reproduction (in Chinese)

Usage

Environment

Step by Step with Conda:

conda create -n kan4tsf python=3.10
conda activate kan4tsf
conda install pytorch torchvision torchaudio pytorch-cuda=12.4 -c pytorch -c nvidia
python -m pip install lightning

or you can just:

pip install -r requirements.txt

Code and Data

ETTh1 and ETTm1 can be downloaded within this project, and other datasets can be downloaded from [Baidu Drive].

Running

python train.py -c config/reproduce_conf/RMoK/ETTh1_96for96.py

Cite

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

@inproceedings{han2023are,
  title={Are KANs Effective for Time Series Forecasting?},
  author={Xiao Han, Xinfeng Zhang, Yiling Wu, Zhenduo Zhang and Zhe Wu},
  booktitle={arXiv},
  year={2024},
}