W-Transformers : A Wavelet-based Transformer Framework for Univariate Time Series Forecasting

W-Transformer is a new framework to forecast univariate time series. To know more about the achievement of W-Transformers or contact us for more information, you can find the full paper here.

Installation

git clone https://github.com/CapWidow/W-Transformer.git
cd w-transformer
pip install -r requirement.txt

Usage

python WTransformer.py

Forcasting horizon

In the code, you can change the variable :

nforecastList = [288,576]

this list are the forecasting horizon that you want. (Here for the NetworkAnalytics.csv)

Our recommandation are:

For Daily data

nforecastList = [30,90]

For Weekly data

nforecastList = [26,52]

For Monthly data

nforecastList = [12,24]

Data

The data you want to use should be in the folder refer in the variable

 path = 'C:/Users/Admin/ResearchW/MODWT/Data/Data_tmp'

Change the path variable in the code by your folder where your data are located.

 path = 'PATH/TO/YOUR/DATA/FOLDER/'

You can put multiple csv in this folder and they will all be execute. Please note that the column name of the value should be named 'Cases' as the execution is automated with this name.

Contributing

Pull requests are welcome. For major changes, please open an issue first to discuss what you would like to change.

Cite

If this research help you in your work please cite W-Transformers :

@article{sasal2022w,
  title={W-Transformers: A Wavelet-based Transformer Framework for Univariate Time Series Forecasting},
  author={Sasal, Lena and Chakraborty, Tanujit and Hadid, Abdenour},
  journal={arXiv preprint arXiv:2209.03945},
  year={2022}
}