metaforecast is a Python package for time series forecasting using meta-learning and data-centric techniques.
This package implements various techniques to improve forecasting accuracy based on dynamic model combination, data augmentation, and adaptive learning, building upon Nixtla’s awesome ecosystem of state-of-the-art forecasting methods.
metaforecast currently consists of three main modules:
- Dynamic Ensembles: Combining multiple models with adaptive ensemble techniques.
- Synthetic Time Series Generation: Creating realistic synthetic time series data for robust model training and testing. Includes a special callback for online data augmentation.
- Long-Horizon Meta-Learning: Instance-based meta-learning for multi-step forecasting.
You can install metaforecast using pip:
pip install metaforecastTo install metaforecast from source, clone the repository and run the following command:
git clone https://github.com/vcerqueira/metaforecast
pip install -e metaforecast
cd metaforecast
pre-commit installCheck the documentation for the API reference and module descriptions. You can get started with a few tutorials.
metaforecast is in the early stages of development. The codebase may undergo significant changes. If you encounter any issues, please report them in GitHub Issues