sktime/pytorch-forecasting
🏗️ Frameworksktime
Time series forecasting made easy with PyTorch, featuring state-of-the-art models.
sktime/pytorch-forecasting simplifies time series forecasting by offering a consistent API for PyTorch-based deep learning models. It includes pre-built models such as N-BEATS, DeepAR, Temporal Fusion Transformer, and more. The library leverages PyTorch Lightning for training, enabling easy GPU support, early stopping, and logging. It also provides utilities for data preprocessing, backtesting, and interpretation. The project is part of the sktime ecosystem, aiming to make forecasting accessible to practitioners.
💡Highlights
- ├─Unified API for PyTorch forecasting models
- ├─Includes N-BEATS, DeepAR, TFT
- └─GPU training with PyTorch Lightning
🎯For
- ├─Data scientists
- ├─Machine learning engineers
- └─Researchers