Describe fundamental principles of machine learning on Azure →mediumMultiple ChoiceObjective-mapped

AI-900 Practice Question: Describe fundamental principles of machine learning on Azure

What is 'cross-validation' and when should it be used in machine learning?

⚠ Common exam trap

Watch out — candidates often confuse cross-validation with simple train/test splitting or model comparison, but the key is recognizing cross-validation as a repeated resampling method to obtain a reliable performance estimate, not a one-time validation or inter-annotator agreement check.

Answer choices

Why each option matters

Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.

Correct answer & explanation

✓

Dividing data into k folds and training k times to get a more reliable performance estimate

Cross-validation is a resampling technique used to evaluate machine learning models by partitioning the original dataset into k equal-sized folds. The model is trained on k-1 folds and validated on the remaining fold, repeating this process k times so each fold serves as the validation set once. This provides a more robust and less biased estimate of model performance compared to a single train-test split, especially when data is limited.

Answer analysis

Option-by-option breakdown

For each option: why learners choose it and why it is or isn't the right answer here.

✗
Validating that a model works correctly across different Azure regions
Why it's wrong here
Validating that a model works correctly across Azure regions concerns deployment operations—for example, verifying that endpoints respond, latency meets SLAs, and data residency complies with regional policies. This is infrastructure and roll-out testing, not a statistical technique for estimating model generalization. Cross-validation instead partitions a single labeled dataset into training and validation folds to assess predictive performance.
✓
Dividing data into k folds and training k times to get a more reliable performance estimate
Why this is correct
K-fold cross-validation divides the dataset into k equally sized, disjoint folds. The model is trained k times, each time holding out a different fold for validation and using the remaining k-1 folds for training, then the k validation scores are averaged to produce a more stable performance estimate. Because every sample appears in a validation fold exactly once, this reduces the variance of the estimate and uses limited data far more efficiently than a single train/test split.
✗
Comparing two different models' predictions on the same test set
Why it's wrong here
Comparing the predictions of two different models on the same static test set is a direct evaluation approach, often visualized with confusion matrices, ROC curves, or metric differences like accuracy and F1. This is model selection or hypothesis testing, but it does not involve resampling the dataset into multiple training/validation splits. Cross-validation could be used to make such a comparison more robust, but the action of comparing on a single test set is distinct from the cross-validation procedure itself.
✗
Checking whether training labels are consistent across different human annotators
Why it's wrong here
Checking that different human annotators assign consistent labels is inter-annotator agreement analysis, often measured with Cohen's kappa or Fleiss' kappa. This targets label noise and annotation guideline clarity, which are pre-modeling data-quality concerns. It is unrelated to cross-validation, which is a resampling procedure performed after labels are finalized to evaluate a model's expected prediction error.

Go deeper

AI-900 Practice Question: Describe fundamental principles of machine learning on Azure

Why each option matters

Option-by-option breakdown

Related to this question