/compas-analysis

Data and analysis for 'Machine Bias'
  1. Jupyter Notebook 99.1%
  2. Python 0.9%
Jupyter Notebook Python

Clone with HTTPS

Use Git or checkout with SVN using the web URL.

Open in Desktop Download ZIP
Find file
Switch branches/tags
Nothing to show
Latest commit bafff5d Jun 13, 2017 @thejefflarson thejefflarson committed on GitHub Merge pull request #4 from mroswell/patch-1 
Typo fix
Permalink
Failed to load latest commit information.
.gitignore this->sky May 22, 2016
Compas Analysis.ipynb Typo fix Jun 13, 2017
Cox with interaction term and independent variables.ipynb update Jul 29, 2016
README add in links to main story and methodology. May 22, 2016
compas-scores-raw.csv raw foia data Jun 8, 2016
compas-scores-two-years-violent.csv this->sky May 22, 2016
compas-scores-two-years.csv this->sky May 22, 2016
compas-scores.csv this->sky May 22, 2016
compas.db this->sky May 22, 2016
cox-parsed.csv this->sky May 22, 2016
cox-violent-parsed.csv this->sky May 22, 2016
truth_tables.py update Jul 29, 2016

README 

                  Low  High
                 +---------+
Didn't Reoffend  |____|____|
Reoffended       |    |    |
                 +---------+


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/

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

Main Dataset:
compas.db - a sqlite3 database containing criminal history, jail and prison time, demographics and COMPAS risk scores for defendants from Broward County.

Other files as needed for the analysis.