What is test data used for?

• A. Data used to train the model
• B. Data used for model deployment
• C. Data used for UI testing
• D. Data used to evaluate model performance on unseen examples

Answer: (D)

Evaluating how a model generalizes to new, unseen data. After training, you set aside a separate test set that the model has not seen during learning and use it to measure performance with metrics like accuracy or F1 score. This provides an unbiased estimate of how it will perform in production on real data. The test set is distinct from training data, which is used to teach the model, and from validation data, which is used during development to tune parameters. So test data is precisely the data used to evaluate performance on unseen examples.