Predicting F1 Race Strategies using ML.NET
summary
Attempts to predict F1 stint lengths using ML.NET regression with initial ~230 manually entered rows then moved to an Ergast/Python pipeline producing 4,000+ rows total (now ~6,000), testing multiple regressors and reporting low but improving R². Categorical features get one‑hot encoding, causing dimensional explosion; mitigation is replacing track names with numerical Pirelli track metrics. Work in progress adds a lap‑time degradation model and pit‑stop time modeling; results are imperfect but progressively better.
AI-written summary · about
chapters (15)
- 0:00F1 strategy prediction intro
- 1:10Why ML.NET for ML
- 3:02ML.NET training workflow
- 4:49ML.NET samples start
- 5:55F1 race strategy basics
- 8:45Tire compounds focus
- 11:54Tire lap regression
- 12:58Categorical data challenges
- 15:24Model updates with new data
- 17:09Testing on Spanish GP
- 18:57Python API data collection
- 20:442022 regulation changes
- 22:53Tire performance curves
- 24:59Future directions
- 26:11Q&A
related talks
45mMachine Learning for .NET developers
26mPredicting the future (of equipment) using ML.NET
30mShift your ML model into overdrive with AWS DeepRacer analysis tools
56mMachine Learning for .NET developers
01h 02mMachine Learning with F#, Redux
20mInnovating at speed
28mShift your ML model into overdrive with AWS DeepRacer analysis tools
28mShift your ML model into overdrive with AWS DeepRacer analysis tools