F1 Safety Prediction ML Model

Formula 1 is a sport where split-second variables—from sudden weather shifts to unexpected traffic—can instantly turn a routine lap into a dangerous situation. Witnessing how quickly a race can change inspired us to tackle the Ollon Track challenge, driven by a desire to see if we could build a system smart enough to spot elevated track risks before race control even makes the call.

Building it wasn't easy; our biggest hurdles were diving into messy, unlabelled motorsport data and figuring out how to handle severe class imbalance where safety events are naturally rare. We had to carefully engineer our own proxy targets and work through the nuances of tuning our model to catch those subtle warning signs. Through the process, we learned invaluable lessons about the power of sequential decision trees, the importance of robust feature engineering, and how to successfully translate complex race telemetry into an intuitive, real-time decision-support tool.