precisions

Precision refers to the degree to which repeated measurements or computations yield similar results under unchanged conditions. It is distinct from accuracy, which concerns how close a result is to a true or accepted value. In metrology and experimental work, precision is often described by the spread of measurements, using statistics such as standard deviation or range. Concepts such as repeatability (same method, same conditions) and reproducibility (different methods or operators, possibly over time) are used to characterize precision. A key idea is that high precision does not guarantee high accuracy; measurements can be consistently biased.

In statistics, precision can denote the inverse of the variance of an estimator, reflecting how tightly the estimator concentrates around the parameter. A more precise estimator has smaller uncertainty and narrower confidence intervals.

In computing, precision describes the numeric accuracy of a representation. Floating-point systems typically offer single precision (about 7 decimal digits) and double precision (about 15–16 digits). Finite precision introduces rounding errors and can affect numerical stability, especially in iterative computations. Some applications use extended or arbitrary precision to control these errors.

In information retrieval and machine learning, precision is a performance metric defined as the ratio of true positives to the sum of true positives and false positives. It measures the correctness of retrieved or predicted items, and is commonly considered alongside recall, which measures completeness. The F1 score combines precision and recall into a single metric.

Across disciplines, precision remains a central concept relating to consistency, exactness, and reliability of measurements, representations, or results, with domain-specific definitions and implications.