How big is each validation fold in k-fold CV?

Question

You have 500 observations and run 10-fold cross-validation: the data is split into 10 equal folds, and each fold takes a turn as the held-out validation set while the model trains on the rest. How many observations are in each held-out validation fold?

Accepted Answer

In k-fold cross-validation the data is divided into k folds of roughly equal size. With 500 observations and k = 10: \frac{500}{10} = 50 	ext{ observations per fold}. Each of the 10 folds serves once as the validation set (with 50 points), while the model trains on the other 9 folds (450 points). Averaging the 10 validation errors gives the cross-validated performance estimate. Takeaway. Each fold holds n/k of the data; with n = 500 and k = 10 that is 50 observations validated at a time.

How big is each validation fold in k-fold CV?

Your answer

Discussion

Learn the concepts

Related questions