Frozen Lake Q-Learning (Implementing Q-Learning in code with Python)

A simple Q-learning implementation to solve the FrozenLake environment from OpenAI Gymnasium.

What it does

The agent learns to navigate a frozen lake to reach a goal while avoiding holes using Q-learning reinforcement learning algorithm.

Requirements

pip install numpy gymnasium

How to run

python main.py

Parameters

• Episodes: 10,000 training episodes
• Learning rate: 0.1
• Discount rate: 0.99
• Exploration rate: Decays from 1.0 to 0.01

Output

• Initial Q-table (all zeros)

• Average reward per 1000 episodes during training

• Final trained Q-table

The agent starts with random exploration and gradually learns optimal actions through trial and error.

Credits

Code adapted from: DeepLizard Q-Learning Tutorial