apress/applied-reinforcement-learning-w-python

Source Code for 'Applied Reinforcement Learning with Python' by Taweh Beysolow

Apress Source Code

This repository accompanies Applied Reinforcement Learning with Python by Taweh Beysolow (Apress, 2019).

Download the files as a zip using the green button, or clone the repository to your machine using Git.

Releases

Release v1.0 corresponds to the code in the published book, without corrections or updates.

Contributions

See the file Contributing.md for more information on how you can contribute to this repository.

applied-reinforcement-learning-w-python's Issues

Warning for Model(input - missing an "s"

C:\Users\markUser\py3\rlEg\neural_networks\policy_gradient_utilities.py:53: UserWarning: Update your Model call to the Keras 2 API: Model(outputs=Tensor("de..., inputs=[<tf.Tenso...)
model_prediction = Model(input=[input_layer], outputs=output_layer)

Looks like the Model(input= ought to be Model(inputs=
^
Note the addition of the "s"
Seems all other lines in this model have the "s" with the exception of line 53.
Adding the "s" removes the warning when I run. Hope this is helpful.

PS - enjoying the book thus far.
Buytore

The code for chapter 4 has something wrong

I figure out that in the market_making_example.py ,line 225 has a bug,it will cause the generator always run the reset(),but each reset() will cause the self.position set into "flat",so the training will always get flat,that is wrong.
I fixed it by making my own reset:MyReset() and just copy the reset but delete the self.position='flat',also delete the releated code with next(),this will make the code right,but still ,the predict result is wrong.
def myReset(self):
self._data_generator.rewind()
self._total_reward = 0
self._total_pnl = 0
self._entry_price = 0
self._exit_price = 0
observation = self._get_observation()
self.state_shape = observation.shape
return observation

AttributeError: 'generator' object has no attribute 'next'

When trying to run generator to test for external csv files, the error is raised : AttributeError: 'generator' object has no attribute 'next' .
The same error is raised when trying to use WavySignal to generate random data