What is model generalization and how is it measured?

Updated May 17, 2026

Short answer

Generalization is a model’s ability to perform well on unseen data.

Deep explanation

Generalization measures how well a trained model applies learned patterns to new, unseen data. It is evaluated using test sets, cross-validation, and metrics like accuracy, F1-score, or RMSE. A model that memorizes training data (overfitting) has poor generalization. Techniques like regularization, dropout, and data augmentation improve it.

Real-world example

A speech recognition model working well on new speakers and accents.

Common mistakes

  • Equating high training accuracy with good generalization.

Follow-up questions

  • What is generalization gap?
  • How do you improve generalization?

More Supervised Learning interview questions

View all →