Reproduce “Neural basis expansion analysis with exogenous variables: Forecasting electricity prices with NBEATSx” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Kin G. Olivares, Cristian Challu, Grzegorz Marcjasz, Rafał Weron, Artur Dubrawski
NBEATSx extends the Neural Basis Expansion Analysis (NBEATS) model to integrate time-dependent exogenous variables and provide interpretable signal decomposition for univariate time series forecasting. Applied to electricity price forecasting, it achieves an average improvement of nearly 20% over the original NBEATS and outperforms other benchmarks without sacrificing accuracy for interpretability.
Key insights
- Explicitly integrating exogenous variables into deep learning forecasting models like NBEATS significantly enhances their predictive accuracy.
- It is possible to achieve interpretable time series decomposition (level, trend, seasonality, and covariate effects) with deep learning architectures without sacrificing overall forecasting performance.
- The modular and residual nature of the NBEATS architecture is well-suited for incorporating new information sources and enabling interpretable signal decomposition.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.