This repository contains results for the paper "A Large-Scale Empirical Study of Aligned Time Series Forecasting". This study explores meta-learning and portfolio approaches for model selection in time series forecasting. The repository contains the results of the experiments with 22 local and 13 global models for time series forecasting.
- Polina Pilyugina - Polina.Pilyugina@skoltech.ru
- Svetlana Medvedeva - S.Medvedeva@skoltech.ru
- Kirill Mosievich - kirillmosievich@gmail.com
- Ilya Trofimov - i-tr@yandex.ru
- Alina Kostromina - amkostromina@edu.hse.ru
- Dmitry Simakov - dmitryevsimakov@gmail.com
- Evgeny Burnaev - E.Burnaev@skoltech.ru
input_data/: Directory containing metrics for models, considering horizons of 7, 30, 90, and 365 days.results/: Directory for saving the results of the experiments.
For further information, please refer to the full paper or contact the authors.