You know "precision -- recall" are important metrics used in ML. You also know their order, ie it's not "recall -- precision". But you can never remember how they are defined? Here is a mnemonic.

In "The boy who cried wolf" everyone makes bad predictions. FIRST the boy has low PRECISION, ie low true positives relative to all cases flagged as positive, TP / (TP + FP). THEN the villagers have low RECALL, ie low true positives relative to all actual positives, TP / (TP + FN).

#ml #statistics