Validation function

[wagenrace]

issue #26

• Add functions to have metrics separate from experiments itself.
• Make notebooks into scripts (this was needed to import utils from every place)
• reduce amount of code
• run experiment 5 times to find out if results are constant

I performed the following prior to filing this pull request:

• Tested that my change does not break the analysis pipeline
• Ran a linter through my code
• Update environment dependencies if my code introduces a new package

[gwaybio]

This is great - adding F1 score and confusion matrix code is very important. I made several comments on most files, but I will summarize my main points below:

1. I highly recommend moving away from .xlsx files to .csv or .tsv files. .xlsx files are much heavier weight than what we need for this project.
2. I couldn't tell how the model was being trained with 6 different initializations. Please clarify.
3. We need to make sure you're properly getting credit for these important contributions! I am seeing grey boxes which indicate that the commit is not properly being attributed to your username. Make sure that the email address you're using to make the commits on your local machine is linked to your github account.

[wagenrace]

@gwaygenomics that should be it

[gwaybio]

Nearly there - only a couple minor and optional comments and just some additional cleanup required.

• There are still some confusion_matrix.xlsx files here. Lets replace them with .csv files.
• Please remove the binary file ~$all_scores.xlsx

Otherwise, looks great! 🎉

[gwaybio]

@wagenrace - confirming that you are not waiting on me - just a couple .xlsx files to be replaced with .csv and the one ~$all_scores.xlsx file to be removed.

Then we are good to merge! 🎉