A machine learning practitioner uses 5x2 cross-validation to compare two models. Which statistical test is specifically designed for this comparison procedure?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Here's the deal — c is correct because Dietterich's 5x2 cross-validated paired t-test correctly accounts for the variance and correlation structure of the 5x2 design. A ignores the paired nature of CV folds.
Full explanation below image
Full Explanation
C is correct because Dietterich's 5x2 cross-validated paired t-test correctly accounts for the variance and correlation structure of the 5x2 design. A ignores the paired nature of CV folds. B (McNemar's) tests differences on a single test set. D describes a different approach (single 5-fold, Wilcoxon) not the 5x2 design.