Why use holdout sets

It is important to understand when you would use three datasets (training, validation, and testing) instead of two (training and testing). There is no point in creating an additional dataset split if you are not going to use it.

When should you consider using training, validation, and testing datasets?

Answer the question

50 XP

Possible Answers

When there is a lot of data. Splitting into three sets helps speed up modeling.
press1
When testing parameters, tuning hyper-parameters, or anytime you are frequently evaluating model performance.
press2
Only when you are running regression and not classification models.
press3
Only when you are running classification and not regression models
press4