A reference list of the observational criteria for fairness from Moritzâ€™s slides. View the slides for intuition + pros/cons.

## Setup

: features of an individual

: sensitive attribute (race, gender)

: classifier mapping and to some prediction

: actual outcome

Assume are binary 0/1 variables.

## Criteria

For all groups

**Demographic parity**:

**Accuracy parity**:

### Positive rate parity

**True positive parity (TPP)**:

**False positive parity (FPP)**:

**Positive rate parity/equalized odds**:

TPP+FPP

### Predictive value parity

**Positive predictive value parity**:

**Negative predictive value parity**:

**Predictive value parity**:

Positive + negative predictive value parity