Application of machine learning for predicting Formula 1 race results

With the growing popularity of motorsports and the increasing availability of large telemetry datasets, machine learning techniques to predict Formula 1 race results have become particularly well-justified. The purpose of this study is to build an ensemble learning model using Support Vector Machine, Gradient Boosting and Random Forest for race result classification. The Optuna library was used for hyperparameter optimisation. The models are based on historical data on races and drivers. The model achieved an F1 score of 77.83% ± 4.18% (macro-averaged) and an accuracy of 80.22% ± 4.69% on the validation set. The results confirm the effectiveness of the applied methods and highlight the significant impact of telemetry data on prediction quality. Ensemble learning can serve as a valuable tool to support Formula 1 race strategies.