VEST: Vector of Statistics for Time Series

This repository contains the original tools for automatic feature engineering for time series forecasting problems.

---

⚠️ Deprecation Warning

This repository is no longer maintained. A newer, more active Python version is available and should be used for current projects:

• vcerqueira/experiments-vest

---

About VEST

VEST (Vector of Statistics from Time Series) provides a framework for automatic feature engineering. It is designed to extract meaningful statistical representations from time series data to improve the performance of forecasting models.

This work is the complementary implementation of the paper: "Vest: Automatic feature engineering for forecasting", published in Machine Learning (2024).

---

Citation

If you find this work useful, please cite the following publication:

@article{cerqueira2024vest,
  title={Vest: Automatic feature engineering for forecasting},
  author={Cerqueira, Vitor and Moniz, Nuno and Soares, Carlos},
  journal={Machine Learning},
  volume={113},
  number={7},
  pages={4523--4545},
  year={2024},
  publisher={Springer}
}