In tasks involving binary classification, positive predictive value (PPV) and negative predictive value (NPV) are respectively the probabilities that a positive or negative prediction is correctly labeled.
We define positive predictive value, also known as precision, as
We define negative predictive value as
PPV and NPV should not be confused with sensitivity and specificity:
Positive predictive value tells us the probability, given a positive prediction, that the prediction is correct. Sensitivity tells us the probability, given an actual positive outcome, that we will predict it as such.
Negative predictive value tells us the probability, given a negative prediction, that the prediction is correct. Specificity tells us the probability, given an actual negative outcome, that we will predict it as such.