Reproduce “A Benchmark for Electrical Load Forecasting Across Grid Levels: Time-Series Transformers Outperform Established Methods” 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
Matthias Hertel, Sebastian Pütz, Jonathan Kolar, Benjamin Schäfer, Ralf Mikut, Veit Hagenmeyer
The authors benchmark load forecasting across a control area, 200 feeders, and 287 customer series, using past load and weather and calendar covariates to predict 96 hourly values. Transformer-based methods have lower error than non-Transformer methods across datasets. On TransnetBW, the standard Transformer reports 143.6 MW MAE versus CNN’s 156.5 MW; Chronos-2 leads on clients, and operational errors may be higher.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.