change suggestions

[laurakn]

Hi Amanda,

I wanted to work on my python skills before I headed out to that internship, so I figured I would do that by working on understanding and cleaning up the COMPAS code. Feel free to accept or reject any changes.

A couple maybe useful changes are there are three methods of choosing how to index the data, so we can look at it with different splits more easily. I also have it so you can just add and remove predictive features at the beginning, cleaned up some of the longer functions and used the logistic regression package.

Everything is a pandas df throughout, so it's easy to compare what the raw and debiased data looks like if one wanted to. I've also used column names instead of positions so it's maybe a little easier to follow.

If the paper is accepted, we should work on making it look super nice and add a link so others check out your analysis!

Great job on putting it all together in such a short time!

Best,

Laura

[Amandarg]

Hey Laura,

Thanks! I think all the changes you made are super helpful, especially adding/removing predictive features and just cleaning it up in general. And yes, I agree we can work on making it super nice if it gets accepted. =)

Amanda