A coworker asked if anyone was interested in attempting to train a model to play tic tac toe. It sounded fun, so this project was my quick (~four hour) attempt.
My plan was
- Create a training set by playing many games of tic tac toe against an opponent whose technique was pure randomness
- Use this training set to train a model, we specifically used a Decision Tree Classifier here
- Have the model play against me, post training
The majority of my time was setting up the code to perform the game, handle the rules of tic tac toe and write back the games for a training set. It was a fun project and I'm happy I did it