Cross-Validation

Definition

Cross-validation is a process for creating a distribution of pairs of training and test sets out of a single data set. In cross validation the data are partitioned into k subsets, S₁… S _k , each called a fold. The folds are usually of approximately the same size. The learning algorithm is then applied k times, for i = 1 to k, each time using the union of all subsets other than S _i as the training set and using S _i as the test set.

Cross-References

• Algorithm Evaluation

• Leave-One-Out Cross-Validation