False Positive Rate
noun
say it: “FAWLS POZ-ih-tiv RAYT” /ˈfɔls ˈpɑzətɪv ˈreɪt/
Of all the cases that are really negative, the share a test or model wrongly marks as positive.
What is the false positive rate?
The false positive rate is, of the examples that really are negative, the share the model wrongly said yes to: FP / (FP + TN). In a fairness check, it's the share of a group's unqualified people the screen picked anyway; equalized odds asks for it to match across groups, along with recall.
fp, tn = 6, 54 # unqualified applicants invited, and not invited
print(fp / (fp + tn)) # 0.1
In a sentence
- “The screen's false positive rate was 0.1 in North and 0.25 in South.”
- “Equalized odds compares the false positive rate of each group.”
Level: Journeyman · More AI Basics terms at this level
Updated