Accuracy is a measure of how close a given measurement is to its true value; precision is how close measurements are to each other. Note, however, that the meaning of “precision” depends on context.
In the context above, when picturing a dart board:
- Accuracy is how close to the bullseye the average throw is, and
- Precision is how tight the grouping is.
Precision is ambiguous
Unfortunately, the word “precision” is also used in relation to recall when discussing information retrieval tasks. The definition of precision in that context has no deep connection; it’s just an unfortunate choice of name.
Accuracy in classification
In classification, accuracy is the proportion of classifications that are correct:
This definition applies both to binary and multi-class classification tasks.