Linked from

The 15 pages that link to Cross-validation, each with the reason it gives.

• Machine learningRelated: It helps assess generalization when a single train-test split is unreliable.

• OverfittingRelated: Repeated held-out estimates help detect overfitting and select model settings.

• RegularizationRelated: It is widely used to choose regularization strength without judging fit on training data alone.

• ResidualRelated: Out-of-sample residuals measure prediction errors beyond the fitted observations.

• Parameter estimationRelated: It assesses generalization rather than simply how well parameters fit the training observations.

• Bootstrap methodCompared with: It also reuses observed data, but targets prediction error rather than sampling uncertainty.

• Predictive validityRelated: Testing a prediction rule on new cases helps reveal whether its validity generalizes.

• ClassificationRelated: It tests whether classification performance generalizes beyond the training examples.

• Data miningRelated: It helps distinguish genuine predictive patterns from results that fit only the training data.

• Ridge regressionRelated: It is commonly used to choose ridge regression's penalty strength.

• Data scienceRelated: It tests whether a fitted model generalizes beyond its training data.

• Minimum description lengthCompared with: It selects models through held-out prediction rather than total encoded description length.

• Coefficient of determinationRelated: Cross-validated R² assesses prediction beyond the data used to estimate coefficients.

• ChemometricsRelated: It helps detect overfitting when chemometric models are built from limited measurements.

• Leave-one-subject-out cross-validationNarrower topic: This method is a cross-validation design in which each fold contains one subject.