I have a very basic question.
1) When is it recommended to hold part of the data for validation and when is it unnecessary? For example, when can we say it is better to have 80% training, 10% validating and 10% testing split and when can we say it is enough to have a simple 80% training and 20% testing split?
2) Also, does using K-Cross Validation go with the simple split (training-testing)?
1) In 80-10-10 scheme, 80% is for training, 10% is for validation and 10% is for testing. Validation set required to search for the optimal hyperparameters.
For models having no hyperparameters, it doesn't do much good to use a validation set (although, it is still useful in determining when to stop the training of the model using early stop). In such situation, one might just keep 80% as training set and 20% as testing set.
2) Yes, K fold CV can be used with simple split.
