0 alternative definitions
A technique for evaluating models by dividing a dataset into K subsets to iteratively train and test the models, and then averaging the results to estimate the overall performance.