Skip to content

Latest commit

 

History

History
16 lines (11 loc) · 814 Bytes

README.md

File metadata and controls

16 lines (11 loc) · 814 Bytes

COMPAS analysis in Python

We aim to redo Propublica analysis in Python. This project was created for the MIT Structure and Interpretation of Deep Networks IAP class.

For more detailed information about the COMPAS dataset and the derivations, I encourage you to check out the Propublica repo.


This repository contains a Jupyter notebook and data for the ProPublica story "Machine Bias."

Story: https://www.propublica.org/article/machine-bias-risk-assessments-in-criminal-sentencing/

Methodology: https://www.propublica.org/article/how-we-analyzed-the-compas-recidivism-algorithm/

Original Notebook in R (you'll probably want to follow along in the methodology): https://github.com/propublica/compas-analysis/blob/master/Compas%20Analysis.ipynb