This is a cryptocurrency price prediction program using Python's keras module and Tensorflow.
To execute, do:
$ python main.py [-s, --seed]--s or --seed is an int of the seed desired to use, if ignored, it defaults to 123
After some time, this should display:
Using [Tensorflow/Theano] backend.
prediction coins by Lachi Balabanski
The backend used by the program is whichever you have installed or keras prefers, namely Tensorflow and Theano.
Next, the program will prompt you for the cryptocurrency abbreviation and the length of time to train from. It is important to note: The larger dataset the program has, the better predictions it will be able to make.
The program will then show a matplotlib graph of the history of the coin that it is training from. This is a sample of the bitcoin graph for a month:
Following that, the program will display another matplotlib generated graph of its predictions, similar to the one above, but with "Predictions" instead of "History" in its title.
It is important to note that upon the first use, a new folder called "__pycache__" will be created. This folder contains a compressed version of the genData.py file.
Secondly, if the program displays something along the lines of:
Your CPU supports instructions that this TensorFlow binary was not compiled to use:
This means that your CPU has special operators that tensorflow could, but did not, use. If you would like to learn more, it is much better answered by this stackoverflow response