This is a follow-up question to another post on SE AI that asked to distinguish lazy and eager learning. One answer said that lazy learners do not require training and do all of the computation during inference.
In my limited experience, it seems that some sort of validation (e.g. holdout validation, cross-validation) is considered standard practice, but the answer seems to imply otherwise for lazy learners. Is it correct to say that since training is not required, the usual train-test split/validations are not necessary for lazy learning?
