Artificial Intelligence · Model evaluation
Why do we keep test data separate from training data?
Student 1
Why do we keep test data separate from training data?
Faculty
A separate test set measures how well the model handles examples it did not train on. Keep it out of model fitting and tuning, then use it for a final evaluation.
Student 2
Which data should I use to compare two models?
Faculty
Use validation data or cross-validation to compare models. Save the test set for evaluating the final choice.