|
|
Welcome to my Player vs Computer American Checkers game! This project demonstrates my software engineering and machine learning skills. The game incorporates various technologies, data manipulation, and artificial intelligence to create an engaging and challenging experience.
Make sure you have python installed on your machine. Then, from within the Checkers directory, install requirements
pip install -r requirements.txt
From within the Checkers directory, run the following command (might take approx. 1 minute to load on the first time)
python checkers_game/main.py
I acquired a dataset of 20,000+ American Checkers games, manipulated it to extract move records, and analyzed board states, resulting in 1 million+ samples.
Using a 2D matrix representation and leveraging an external library, I parsed and analyzed each game state to extract 11 features.
These processes occurred in the "data_manipulation" folder.
In the "agent_creation" folder, Jupyter Notebook was used to test various neural network and machine learning models on the dataset to identify the best-performing model.
The chosen model was integrated into the "checkers_game" folder, responsible for executing all game logic and gameplay, including the user interface and the Checkers bot's move handler.
In addition to the neural network model, I implemented a Minimax algorithm with ⍺-β pruning, that allows the Checkers bot to explore move trees up to depth 5 in hard mode, providing a formidable opponent.
Throughout the project, I utilized various technologies, including:
- PyTorch
- scikit-learn
- Jupyter Notebook
- numpy
- pandas
- pygame
And more...
Feel free to explore the codebase, and reach out if you have any questions or feedback!