Simulation to predict timeserie of cryptocurrency using binance api for the dataset. It's base on gymnasium template for simulation. Contain some data augmentation base on approximate entropy, fractal dimension(hurst dimension), lag value and relative change.
It is a simulation to use Deep Reinforcement learning for stochastic time series with a discret action space and a 3D continuous observation space.
The script register_gym_simulation.py will register the envs ( https://gymnasium.farama.org/tutorials/gymnasium_basics/environment_creation/ ) instead of the manual process.
One will find pseudocode use case example in each file.