What is the role of validation error in bias-variance tradeoff?
Updated May 15, 2026
Short answer
Validation error helps estimate generalization performance and identify optimal bias-variance balance.
Deep explanation
Validation error is used to evaluate how well a model generalizes to unseen data. It helps detect overfitting (high variance) and underfitting (high bias). By comparing training and validation errors, practitioners can adjust model complexity and hyperparameters.
Real-world example
Tuning a recommendation model using cross-validation to avoid overfitting to historical user behavior.
Common mistakes
- Using training accuracy instead of validation accuracy for evaluation.
Follow-up questions
- What is cross-validation?
- Why is test set different from validation set?