README.md

Notebook Instructions

To run the prediction program, ensure that you are on a terminal such as Git Bash with a virtual Python environment already activated.

Then issue the following commands:

git clone https://github.com/cybertraining-dsc/su21-reu-361.git
cd su21-reu-361/project/code
pip install -r requirements.txt

You may create a new config file (ending in .yaml) and edit the beginning cells in yfinance-lstm-all-figures.ipynb to use your specific config file. This config file produces log files, which are necessary to run the next analysis notebook called yfinance-lstm-analysis-final.ipynb. If desired, you may also change the list of cryptocurrency tickers that are to be predicted.

Issue the following after making your own config file and editing the notebook to use your config file:

jupyter nbconvert --to notebook --inplace --execute yfinance-lstm-all-figures.ipynb --ExecutePreprocessor.timeout=600

If you would also like to run the analysis script to produce the figures, you must also change the filename variable in the yfinance-lstm-analysis-final.ipynb notebook to use your specific log file. Then issue the following:

jupyter nbconvert --to notebook --inplace --execute yfinance-lstm-analysis-final.ipynb --ExecutePreprocessor.timeout=600

TODO: add requirements.txt, git clone, set up pyenv, make sure the reader knows this for ease of replication

TODO: add papermill to requirements and create makefile that runs papermill

cp yfinance-lstm.ipynb yfinance-lstm-hostname.ipynb papermill yfinance-lstm-hostname.ipynb

This will produce the new notebook with all the results included An alternative way is to convert the ipynb to python with nbconvert ... some options

You can also run this directly in jupyter-lab

jupyter-lab yfinance-lstm-hostname.ipynb

Please ensure you have downloaded the code folder. GitHub and git makes this easy through cloning of the repo. Cloning is impossible without first installing Git. Alternatively one can download the repository as a ZIP and extract it by clicking the green Code button on GitHub. Also, an IDE must be installed, such as PyCharm or Visual Studio Code.

Once downloaded, open the yfinance-lstm.ipynb in an IDE, preferably in a Python virtual environment. Tutorials on how to start a virtual environment are in the tutorials folder of the repo. Then, ensure all the Python modules that are imported within the first cell of yfinance-lstm.ipynb are installed with pip (preferably within virtual environment).

Then, click Run All. There will be a prompt to input the ticker of the cryptocurrency. Enter it, and the program should do the rest.

However, near the end of the program, there is a line that zooms into the axes of the graph. The parameters should be manually changed so that it zooms into a point of interest. Because cryptocurrencies have wildly varying prices, this must be changed on a case-by-case basis.

NOTE: In order to generate the sequential model diagram on Windows, graphviz must be installed. We recommend that you install graphviz via chocolatey. First install chocolatey and then run choco install graphviz.