Core Concepts

Understand the foundations of short-term energy forecasting and the design decisions behind OpenSTEF’s machine learning framework.

Forecasting Basics (Short-Term Forecasting Basics)

What short-term energy forecasting is, why it matters for grid operations, and how it differs from other prediction tasks.

Quantiles and Confidence (Probabilistic Forecasts and Quantiles)

How OpenSTEF produces probabilistic forecasts with uncertainty bands, and how to interpret quantile predictions.

Feature Engineering (Feature Engineering for Energy Forecasting)

The domain-specific predictors OpenSTEF uses—weather data, time features, lag variables—and how they improve forecast accuracy.

Reliability and Fallback (Reliability and Fallback Strategies)

How OpenSTEF handles production failures gracefully: fallback strategies for missing data, failed models, and degraded inputs.

Meta-Ensembles (Ensemble Forecasting)

Why combining multiple models outperforms any single approach, and how the ensemble architecture works.

Component Splitting (Energy Component Splitting)

Decomposing aggregate load measurements into constituent energy components like solar, wind, and base load.