Skip to content

Warning You're not viewing the latest version. Click here to go to latest.

skforecast

Time series forecasting with scikit-learn regressors.

Skforecast is a python library that eases using scikit-learn regressors as multi-step forecasters. It also works with any regressor compatible with the scikit-learn API (XGBoost, LightGBM, Ranger...).

Installation

1
pip install skforecast

Specific version:

1
pip install git+https://github.com/JoaquinAmatRodrigo/skforecast@v0.1.9

Latest (unstable):

1
pip install git+https://github.com/JoaquinAmatRodrigo/skforecast#master

The most common error when importing the library is:

'cannot import name 'mean_absolute_percentage_error' from 'sklearn.metrics'.

This is because the scikit-learn installation is lower than 0.24. Try to upgrade scikit-learn with

pip install scikit-learn==0.24

Dependencies

1
2
3
4
5
6
python>=3.7.1
numpy>=1.20.1
pandas>=1.2.2
tqdm>=4.57.0
scikit-learn>=0.24
statsmodels>=0.12.2

Features

  • Create recursive autoregressive forecasters from any scikit-learn regressor
  • Create multi-output autoregressive forecasters from any scikit-learn regressor
  • Grid search to find optimal hyperparameters
  • Grid search to find optimal lags (predictors)
  • Include exogenous variables as predictors
  • Include custom predictors (rolling mean, rolling variance ...)
  • Backtesting
  • Prediction interval estimated by bootstrapping
  • Get predictor importance

References

  • Hyndman, R.J., & Athanasopoulos, G. (2018) Forecasting: principles and practice, 2nd edition, OTexts: Melbourne, Australia

  • Time Series Analysis and Forecasting with ADAM Ivan Svetunkov

  • Python for Finance: Mastering Data-Driven Finance

Licence

joaquinAmatRodrigo/skforecast is licensed under the MIT License, a short and simple permissive license with conditions only requiring preservation of copyright and license notices. Licensed works, modifications, and larger works may be distributed under different terms and without source code.