What are precision and recall?

In plain English

Precision: of everything the model flagged, how much was really a hit? Recall: of all the real cases out there, how many did the model find? Usually you trade one for the other: catch more, and you also raise more false alarms.

An example

Say 10 in every 1,000 card payments are fraud. A model that never flags anything is 99% accurate, and catches no fraud at all: its recall is zero. Accuracy alone would call it excellent.

Watch out

Which one matters more depends on the cost of each mistake. A cancer screen should favour recall, because missing a case is worst. A spam filter should favour precision, because losing a real email is worse than seeing some spam. Try it in how a machine learning model is built.