, a model
's quality of producing predictions
close to the theoretical "true" result . Distinct from the statistical definition of precision
, which is the model's quality of tending to converge
on similar predictions
for varying samples of the same population.
To illustrate the distinction, if you throw ten darts at the center of a dartboard, and they are scattered loosely around the center, your throws had high accuracy but low precision. If the darts are tightly grouped at a spot on the wall not very close to the dartboard, your throws had high precision but low accuracy.