Advancing AI and ML Techniques for Short-Term Energy Price Forecasting - Benchmarking Deep Learning, Ensemble Methods, and Generative Models for Enhanced Predictive Performance

Topic:

Are you eager to apply cutting-edge artificial intelligence and machine learning methods to solve real-world challenges in the energy sector? ⚡We are seeking a motivated master's student to explore and benchmark innovative approaches for forecasting energy prices.

In this thesis, you will critically evaluate alternative ML and AI techniques—including, but not limited to, deep learning, ensemble models, and generative AI—against our current forecasting solution. Your goal will be to identify, implement, and validate methods that can improve forecast accuracy, robustness, or interpretability.

Goal:

• Perform a thorough literature review, to identify the state-of-the-art methods used in this sector.
• Review and analyze the current ML-based energy price forecasting tool.
• Research and implement alternative ML/AI approaches (e.g., LSTM, transformers, XGBoost, hybrid models, explainable AI).
• Design and execute benchmarking experiments to compare models.
• Analyze results and provide actionable insights for model improvement, document and present your work

Your profile:

• Master Student in Data Science, Quantitative Finance, Electrical Engineering, Mathematics, Physics, or a related quantitative field
• Strong understanding of machine learning and/or artificial intelligence, with hands-on experience in developing and evaluating ML models, as well as working with time series data, data preprocessing, and feature engineering.
• Solid foundation in statistics, probability, and linear algebra.
• Proficiency in Python, including experience with relevant ML libraries such as scikit-learn, pandas, TensorFlow, or PyTorch.
• Ability to formulate hypotheses, design experiments, and interpret results critically. Strong written and verbal communication skills in English.
• Familiarity with energy systems and energy markets is an advantage but not required.

Starting Date: As soon as possible