Ensemble Machine Learning Models for Predicting Formula 1 Winners and Podium Outcomes

This work presents an ensemble machine-learning framework for predicting Formula 1 race winners and podium finishers using publicly available historical data. We evaluate multiple models including Logistic Regression, Random Forest, XGBoost, and a Stacking Ensemble that combines all three. The study demonstrates that competitive predictive performance—0.894 accuracy for podium prediction and 0.938 accuracy for winner prediction—can be achieved using simple features such as grid position, constructor, driver, and season year, without any proprietary telemetry. The paper also includes a case study analyzing the 2021 Abu Dhabi Grand Prix to demonstrate model behavior in a real, high-stakes scenario. The full methodology, datasets, Python scripts, confusion matrices, and reproducible pipeline are included to support transparency and future research.