Skip to content

lecheuklun/f1ml

Repository files navigation

MATLAB F1 Neural Network Strategy Predictor

A neural network that uses supervised learning on past races to predict a driver's finishing position!

The net achieved 81% accuracy (to +/- 2 positions), compared to 57% with polynomial regression.

https://www.hutch.io/our-games/f1-clash/

Neural network

To run code, run neural-network/test7.m on the editor on MATLAB! Requires Deep Learning Toolbox.

Inputs:

  • Track (e.g. Austria, input as 'AUT')
  • Wet/dry race
  • Qualifying position (e.g. P7)
  • Driver (e.g. Raikonnen, input as 'RAI')
  • Chosen strategy (e.g. '32@4@' signifies soft 3 laps, hard 2 laps, hard 4 laps)

Output:

  • Predicted finishing position

Polynomial regression

To run code, run regression/test6.m on the editor.

Inputs:

  • Track (e.g. Vietnam, input as 'VIE')
  • Ensure strategy options (line 62) are correct

About

A neural network that can predict the optimum pit stop strategy for a Formula One race simulator game.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors