Recall which classification metric answers which operational question.
Definition What does precision answer? Of the items the model flagged as positive, how many were truly positive? Use it when false alarms are expensive or reviewer capacity is tight. What does recall answer? Of all true positive items, how many did the model catch? Use it when misses are expensive, such as fraud, safety, or legal complaints. Averaging Macro average or micro average? For imbalanced NLP labels, macro views often reveal failures hidden by micro or accuracy. Objection The model has 94 percent accuracy, so why not ship? Use when a stakeholder anchors on one aggregate number. Your line…
Sign up free — one personalized lesson every day, matched to your role and goals.
Already have an account? Sign in