Precision and Accuracy are two independently computable numbers used to summarize a data set.

Precision is the tightness of the data set. Visually, high precision is a small cluster and low precision is not clustered. You may also know precision as its inverse, Error. Little error means high precision, and vice versa. (See also: significant digits, significant figures) A simple precision calculation would be the magnitude of the difference between the maximum and minimum values in the data set. Another way is using standard deviation. Example: precision of {1, 2, 1.5, 1, 1.75, 2, 1} = |2 - 1| = 1

Accuracy is the closeness of the data set to a true or theoretical value. Visually, high accuracy is centered on the target value and low accuracy is offset from it. One way to calculate this is to subtract the mean of the data set from the target value. Example: accuracy of {1, 2, 1.5, 1, 1.75, 2, 1} with respect to 4/3 (1.333...) = 4/3 - 9.75/7 -.06

Values closer to zero (within the context of the data) indicate higher precision or accuracy.

See also precision and recall.