Model Evaluation · Validation

k-Fold Cross-Validation

Split the data into k folds — each takes its turn as the test set. Averaging beats one lucky split.

folds k 5
data · split into k folds test error per fold
fold
test error
state
press Run
training foldstest foldCV mean
tip: every fold is the test set exactly once — the CV score is their average
kfold.py