Research Focus
Our lab is committed to leveraging cutting-edge machine learning techniques, including LightGBM, Deep Neural Networks, and Temporal Fusion Transformers, to develop accurate, interpretable electricity price forecasting models in Germany. By integrating explainable AI techniques, we aim to uncover insights into market dynamics.
Day-Ahead Price Forecasting
Our hybrid forecasting model predicts electricity prices one day in advance, factoring in renewable energy outputs, historical prices, weather conditions, and various socio-economic indicators. This approach enables us to address the challenges posed by volatile energy markets, improving forecast accuracy and aiding energy stakeholders in decision-making.
Explainable AI
To enhance interpretability, SHAP analysis is applied to reveal each variable’s impact on price changes. SHAP identifies drivers of negative prices, such as high renewable output combined with low demand, improving the model’s transparency and providing valuable insights.
