README.md

metadata

annotations_creators: []
language: []
language_creators: []
license: []
multilinguality: []
pretty_name: compas-recividivsm
size_categories:
  - 1K<n<10K
source_datasets: []
tags:
  - interpretability
  - fairness
task_categories:
  - tabular-classification
task_ids: []

Port of the compas-recidivism dataset from propublica (github here). See details there and use carefully, as there are serious known social impacts and biases present in this dataset.

Basic preprocessing done by the imodels team in this notebook.