Through the combustion of fossil fuels and the resulting emission of carbon dioxide (CO2), the transport sector contributes significantly to the depletion of resources and global climate change. As a result, government regulations are aimed at reducing CO2 emissions from the transportation sector, which also poses new challenges for the automotive industry. To meet these requirements, the transition from vehicles with internal combustion engines (ICEV) to alternative powertrain technologies such as battery electric vehicles (BEV) is being discussed. However, a reliable forecast of future BEV demand is essential for automakers, suppliers and infrastructure operators to plan production, supply chains, charging networks and resource allocation efficiently.
The objective of this thesis is to design, implement and validate a machine learning model that predicts future BEV demand based on historical sales figures, socioeconomic indicators, policy incentive schemes, energy price trends and charging infrastructure data. Various approaches (e.g., Random Forests, Gradient Boosting, LSTM networks) should be compared. Key tasks include data acquisition and preprocessing, feature engineering, model training with hyperparameter optimization, and evaluation using appropriate metrics and scenario analyses.
Successful completion of this project requires proficiency in Python and data-science libraries (pandas, scikit-learn, TensorFlow or PyTorch, etc.) as well as initial experience with machine-learning methods.
If you are interested, please contact Raphael Ginster.