In information retrieval and classification, the fraction of items labeled as positive by a system that are actually true positives; also called positive predictive value.
Degree to which repeated measurements or computations under unchanged conditions yield the same or similar results; commonly associated with low random error or small dispersion of outcomes.