What is cross validation used for in model evaluation?

• A. Increasing dataset size
• B. Estimating model generalization by training and validating on different subsets of the data
• C. Reducing training time
• D. Tuning hyperparameters automatically

Answer: (B)

Cross-validation estimates how well a model generalizes to unseen data by evaluating its performance across multiple train/validation splits of the data. In the common k-fold approach, the data are partitioned into k folds; the model is trained on k−1 folds and tested on the remaining fold, repeated for all folds, and the results are averaged. This provides a more reliable assessment than a single holdout split because it reduces the variability introduced by how the data are split. It helps compare models and tune hyperparameters by giving a robust generalization estimate without needing extra data. It does not inherently increase dataset size or speed up training, and hyperparameter tuning is typically done with additional strategies such as nested cross-validation or separate validation procedures rather than relying on a single cross-validation run.