This project was upgraded from KAN4TSF to EasyTSF (Experiment assistant for your Time-Series Forecasting).
🚩 News (2025.10) Another project: EasyResult. This is an experimental data visualization page that provides data import, model/dataset drag-and-drop, best-value highlighting, and multi-format (CSV and Latex) export functionality.
🚩 News (2025.02) We add MMK, a Multi-layer Mixture-of-KAN model for MTSF.
🚩 News (2024.11) KAN4TSF -> EasyTSF, we will support more time series forecasting models.
🚩 News (2024.09) Model Zoo: RMoK, NLinear, DLinear, RLinear, PatchTST, iTransformer, STID, TimeLLM
🚩 News (2024.09) Introduction and Reproduction (in Chinese)
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 lightningor you can just:
pip install -r requirements.txtETTh1 and ETTm1 can be downloaded within this project, and other datasets can be downloaded from Baidu Drive or Google Drive.
python train.py -c config/reproduce_conf/RMoK/ETTh1_96for96.py python ray_tune.py --config config/reproduce_conf/MMK/MMK_ETTh1_96for96.py --param_space config/reproduce_conf/MMK/MMK_search_space.py If you find this repo useful, please cite our paper:
@inproceedings{han2023are,
title={KAN4TSF: Are KAN and KAN-based models Effective for Time Series Forecasting?},
author={Xiao Han, Xinfeng Zhang, Yiling Wu, Zhenduo Zhang and Zhe Wu},
booktitle={arXiv},
year={2024},
}
or
@inproceedings{10.1145/3746252.3760836,
author = {Han, Xiao and Zhang, Zhenduo and Zhang, Xinfeng and Wu, Yiling and Wu, Zhe},
title = {Mixture-of-KAN for Multivariate Time Series Forecasting},
year = {2025},
isbn = {9798400720406},
publisher = {Association for Computing Machinery},
address = {New York, NY, USA},
url = {https://doi.org/10.1145/3746252.3760836},
doi = {10.1145/3746252.3760836},
pages = {4770–4774},
numpages = {5},
keywords = {kolmogorov-arnold network, multivariate, time series forecasting},
location = {Seoul, Republic of Korea},
series = {CIKM '25}
}