Deep Learning
Precision is a performance metric that defines a classification model's ability to identify only relevant samples.
True
Proposed by @svpino
Card 1 of 259. Answer: True
True
Proposed by @svpino
Card 1 of 259. Answer: True
About · Quizzes · History of AI · To suggest new questions or report an error send me a dm.
True. Precision is the fraction of samples a classifier labels positive that are actually positive: precision = TP / (TP + FP). A high-precision model rarely raises false alarms, which is what "identify only relevant samples" means. Recall instead asks whether it finds all of them.
Example: a spam filter moves 200 emails to the spam folder; 190 are spam and 10 are legitimate, so precision is 190 / 200 = 0.95. If 250 spam emails arrived in total, recall is 190 / 250 = 0.76. The 60 spam emails that reached the inbox do not touch precision at all.
Favor precision when false positives are costly: a real email lost to spam, an innocent account suspended. It is easy to inflate by flagging only the most confident cases, so report it with recall. Interviewers also ask about base rates: with fixed true and false positive rates, precision drops as positives get rarer, while recall stays put. In medicine it is called positive predictive value.