Connect 4 AI Agents 🤖

Welcome to the Connect 4 AI Agents project! This repository contains a collection of AI agents designed to play the classic game of Connect 4. The agents range from simple random decision-makers to more advanced ones like the MiniMax and ML-based agents. Below, you'll find all the necessary information to get started, evaluate the agents, and even play against them!

🚀 Getting Started

To get started, clone this repository and ensure you have Python 3 installed. All commands should be executed from the root directory of the project.

🐍 Python Version and Virtual Environment

This project is developed and tested using Python 3.12.8. To ensure compatibility, it's recommended to use the same version in a virtual environment (.venv).

📦 Install Dependencies

First, install the required dependencies using pip:

pip install -r requirements.txt

🕹️ Play Against the AI

To play Connect 4 against one of the AI agents, run the following command:

python3 -m src.play_game

🧪 Run Tests

We have a suite of tests to ensure everything is working as expected. You can run the tests for each component as follows:

python3 -m src.test.test_game_board
python3 -m src.test.test_mini_max_agent
python3 -m src.test.test_ml_agent
python3 -m src.test.test_random_agent
python3 -m src.test.test_smart_agent

📊 Evaluation

To evaluate the performance of the AI agents, navigate to the src/evaluation/evaluation.py file. Uncomment the relevant lines for the evaluation you want to run, and then execute:

python3 -m src.evaluation

The evaluation results will include graphs and metrics to compare the agents' performance.

📈 Evaluation Graphs

Here are some key graphs generated during the evaluation:

1. Training Loss for ML Agent
   

2. Agent Performance Comparisons
   Random vs. Smart
   Smart vs. Minimax
   Minimax vs. ML
   

📚 Credits

For a list of resources, libraries, and tools used in this project, check out the credits.md file. It includes links and acknowledgments for everything that helped make this project possible.