AI Glossary Term
What does “Validation Set” mean?
Definition
A dataset used during development to tune models and prevent overfitting.
How Validation Set works
Example
While tuning a model, an engineer checks scores on the validation set after each experiment to decide which settings to keep. Once decisions are final, the untouched test set gives the final estimate.
Why it matters
It lets you improve a model without peeking at the final exam. Using the test set for tuning would make results look better than they are.
Common questions
Validation vs test set?
Validation guides development choices; the test set is for a single final check.
What is cross-validation?
Rotating which part of the data is held out, so every example is used for both training and validation across rounds.
How much data?
Often 10 to 20 percent, adjusted to dataset size.