True Positive Rate
Overview
Calculates TP / (TP + FN)
for privileged and unprivileged groups. This is the probability that a positive prediction is truly positive.
Formula
TPR = TP / (TP + FN)
Where:
-
TP (True Positives): The number of positive instances correctly predicted by the model
-
FN (False Negatives): The number of negatives instances incorrectly predicted as positive by the model
Usage
Manually
tpr_priv, tpr_unpriv = true_positive_rate(df, protected_attribute, privileged_group, labels, positive_label, y_true)
print("True Positive Rate for privileged group:", tpr_priv)
print("True Positive Rate for unprivileged group:", tpr_unpriv)
Using Fairness Object
tpr_priv, tpr_unpriv = (fo.compute(true_positive_rate))
Results
True Positive Rate for privileged group: 0.6666666666444444
True Positive Rate for unprivileged group: 0.33333333332777776
These results are obtained by using the input data given in the Create Example Data page under Getting Started
Interpretation
A higher TPR indicates that the positives are being correctly assigned as such.
Updated 9 months ago