False Negative Rate

Overview

Calculates FN / (TP + FN) for privileged and unprivileged groups. This is the rate of false negatives seen out of all the positive events.

Formula

FNR = FN / (TP + FN)

Where:

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

Usage

Manually

fnr_priv, fnr_unpriv = false_negative_rate(df, protected_attribute, privileged_group, labels, positive_label, y_true)

print("False Negative Rate for privileged group:", fnr_priv)
print("False Negative Rate for unprivileged group:", fnr_unpriv)

Using Fairness Object

fnr_priv, fnr_unpriv = (fo.compute(false_negative_rate))

Results

False Negative Rate for privileged group: 0.3333333333222222
False Negative Rate for unprivileged group: 0.6666666666555555

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

Interpretation

A higher FNR indicates a greater likelihood to falsely predict a negative result.