DeCorrelated (large)
This collection contains 22 scenarios (each scenario is a combination of a dataset with a particular selection of sensitive attributes).
1.Folktables ACSPublicCoverage
economicsRAC1P
2.Heart Disease
cardiologysex
3.HMDA
financeapplicant_sex_nameapplicant_race_name_1
4.Stop, Question and Frisk Data
lawSUSPECT_SEXSUSPECT_RACE_DESCRIPTIONSUSPECT_REPORTED_AGE
5.Folktables ACSEmployment (small)
economicsRAC1P
6.Folktables ACSTravelTime
economicsRAC1P
7.COMPAS
lawsexage
8.Folktables ACSIncome (small)
economicsRAC1P
9.COMPAS (2 years)
lawage
10.Communities (unnormalized)
lawpct12-21
11.Arrhythmia
cardiologysex
12.Folktables ACSPublicCoverage (small)
economicsRAC1P
13.COMPAS (2 years, violent)
lawage
14.South German Credit
financeageforeign_worker
15.Dutch
demographyage
16.Folktables ACSMobility (small)
economicsRAC1P
17.Law School (tensorflow)
educationgender
18.German Credit (onehot)
finance<= 25 years
19.Communities
lawracePctAsian
20.Nursery
educationfinance
21.German Credit (numeric)
financeage
22.Chicago Strategic Subject List
lawRACE CODE CD
Example Code
Example code showing how to use this collection in Python: Loading code preview...