Extending Bayesian Decision Trees to Incorporate External Information: A Sprint Kayak and Canoe Case Study

This thesis develops new methods to compare and predict elite kayak and canoe race performances under different environmental conditions. Combining statistical and decision-tree based machine learning, it shows how race times can be adjusted for factors such as wind, temperature, and water conditions, improving accuracy and decision-making. The research also introduces a way to combine model predictions with expert knowledge to estimate how athletes would perform under baseline conditions. These methods provide clearer insights for coaches and athletes and can be applied to other fields where data are complex or incomplete.