Formula 1 used AWS’ machine learning data-analytics service to deliver fans new insights into last weekend’s Australian Grand Prix.
The new Pit Strategy Battle statistic analysed the potential success or failure of an undercut or overcut manoeuvre.
It provided real-time information for F1 teams on the position of two drivers; the predicted gap after pit stops; and the percentage chance of an over take.
This new stat was also delivered direct to race broadcasters, via the Formula 1 International Feed, in order to augment commentary and provide greater insight.
To create the graphic the Formula 1 team gathers data from a number of sources, including information about the position of the drivers on the track, the time the car spends in the pit lane, the time the car is stationary in the pit stop and information about tyre compound. All this information is compared against historic data and fed into a machine learning model, which delivers an overtake probability score.
If a probability score of 50 per cent is forecast, it means the two drivers will be side-by-side.
If a probability of less than 50 per cent is forecast then the chasing driver is predicted to remain behind the leading driver, and if a probability score of more than 50 per cent is a forecast than the chasing driver is predicted to be ahead of the leading driver.
