Each iteration produces one model, and thereby one estimate of generalization performance, for example, one estimate of accuracy. When cross-validation is finished, every example will have been used only once for testing but k–1 times for training.
Iteration
