Colloquium: Casper Dolman (C&O)

Flight delays significantly impact the aviation industry, leading to increased costs, passenger dissatisfaction, and operational challenges. This research focuses on forecasting en route delays for individual flights 90 minutes before departure using supervised machine learning models. By comparing CatBoost and LightGBM algorithms, the study evaluates prediction accuracy and model interpretability, aiming to provide actionable insights for airline dispatchers and pilots. The proposed models incorporate various operational, temporal, and weather-related features, improving prediction performance over existing methods. Results demonstrate the potential of machine learning to enhance delay forecasting, aiding in more efficient flight planning and fuel management

Supervisor: Marta Ribeiro