Folder containing all my projects in Reinforcement Learning.
-
Flappy Bird Game This project implements 2 agents (a Monte Carlo and Sarsa-Lambda) from scratch, using an already-implemented environment of the game Flappy Bird. It works very well on Notebook, even with a dynamic rendering.
Spoiler : the bird flaps well.
- Stock Trading Agent : This project implements a Deep Q Network Agent on 3 Environments. The data is the stock of Apple from 2022 to now (March 2024). The agent has possible 3 actions {buy, sell, hold}, and the state space increases from {current price} to {current price, volume exhanged, rolling average} to {current price, volume exhanged, rolling average, lag-1, lag-5}.
Spoiler : The agent should write 'proud overfitter and portfolio's vanisher' on it's resume.