Temporal Fusion Transformers for Interpretable Multi-horizon Time Series Forecasting
About
Multi-horizon forecasting problems often contain a complex mix of inputs -- including static (i.e. time-invariant) covariates, known future inputs, and other exogenous time series that are only observed historically -- without any prior information on how they interact with the target. While several deep learning models have been proposed for multi-step prediction, they typically comprise black-box models which do not account for the full range of inputs present in common scenarios. In this paper, we introduce the Temporal Fusion Transformer (TFT) -- a novel attention-based architecture which combines high-performance multi-horizon forecasting with interpretable insights into temporal dynamics. To learn temporal relationships at different scales, the TFT utilizes recurrent layers for local processing and interpretable self-attention layers for learning long-term dependencies. The TFT also uses specialized components for the judicious selection of relevant features and a series of gating layers to suppress unnecessary components, enabling high performance in a wide range of regimes. On a variety of real-world datasets, we demonstrate significant performance improvements over existing benchmarks, and showcase three practical interpretability use-cases of TFT.
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Long-term time-series forecasting | ETTh1 | MAE0.519 | 575 | |
| Multivariate Time-series Forecasting | ETTm2 | MSE0.2131 | 539 | |
| Multivariate Time-series Forecasting | Weather | MSE0.1917 | 409 | |
| Anomaly Detection | SMD | -- | 375 | |
| Multivariate Time-series Forecasting | Traffic | MSE0.5921 | 310 | |
| Time Series Forecasting | Weather (test) | -- | 248 | |
| Multivariate Time-series Forecasting | ETTh2 | MSE0.362 | 198 | |
| Multivariate Time-series Forecasting | Electricity | MAE0.3052 | 105 | |
| Time Series Forecasting | ETTh2 | MSE0.217 | 88 | |
| Time Series Forecasting | NP | MSE0.219 | 84 |