What is k-fold cross-validation?

• A. Training multiple models independently with no cross-validation.
• B. Dividing data into k groups and rotating training/testing.
• C. Using k nearest neighbors for classification.
• D. Evaluating on a fixed test set only.

Answer: (B)

Dividing data into k groups and rotating training/testing is k-fold cross-validation. In this approach you split the dataset into k folds of roughly equal size, then for each iteration you train the model on k−1 folds and evaluate it on the remaining fold. You repeat this process until every fold has served as the test set, and then you average the performance across all folds to get a robust estimate of how the model will generalize to new data. This technique reduces the variability that comes from relying on a single train/test split and makes efficient use of the data, which is especially helpful when available data are limited. It also provides a fair basis for comparing models or tuning hyperparameters.

The other descriptions don’t fit this idea: training multiple models independently without cross-validation lacks the validation loop; using k nearest neighbors describes a specific algorithm rather than a validation method; evaluating on a fixed test set only corresponds to a single split, not the iterative testing across multiple folds.