False Omission Rate

Overview

The False Omission Rate (FOR) is a metric used to assess the proportion of false negative predictions among all negative predictions made by a classification model. It quantifies the rate of incorrect negative predictions relative to the total number of negative predictions.

Formula

FOR = P(actual = + | prediction = -) = FN/(TN+FN)

Source

Where:

• FN (False Negatives): The number of positive instances incorrectly predicted as negative by the model.
• TN (True Negatives): The number of negative instances correctly predicted by the model.

Usage

Manually

Python

# Usage example
for_priv, for_unpriv = false_omission_rate(df, protected_attribute, privileged_group, labels, positive_label, y_true)

print("False Omission Rate for privileged group:", for_priv)
print("False Omission Rate for unprivileged group:", for_unpriv)

Using Fairness Object

for_priv,for_unpriv = (fo.compute(false_omission_rate))

Results

False Omission Rate for privileged group: 0.16666666666388888 
False Omission Rate for unprivileged group: 0.799999999984

These results are obtained by using the input data given in the Create Example Data page under Getting Started

Interpretation

A lower FOR value indicates that the model has a lower rate of false negative predictions among all negative predictions, which is desirable. A FOR of 0.0 indicates that there are no false negative predictions made by the model.