cross-validation

n. A statistical technique that splits a sample of data into two or more subsamples in order to provide separate data sets for initial and validation analyses. The theory of cross-validation was developed in response to concerns about testing hypotheses suggested by the data. Cross-validation helps guard against type I error by performing initial analysis on one subset of data while reserving remaining samples for follow-up analysis. Cross-validation techniques are particularly valuable when additional samples are too expensive or dangerous to obtain, or when it is impossible to collect further samples. Cross-validation can be conducted via holdout (also called split- sample method), K-fold, or leave-one-out methodologies.       

- bjm