$2.$ We now review the topic of model validation:

$($a$)$ Explain how validation set approach is implemented. What are its main disad$-$ vantages?

$($b$)$ Explain how the leave$-$one$-$out cross$-$validation is implemented $($LOOCV$).$ How does it improve on validation set approach?

$($c$)$ What is the main disadvantage of LOOCV approach? Hint: Imagine we have $10000$ observations and $20$ variables.

$($d$)$ Could we use validation set & LOOCV for any supervised learning approach? Or only for KNN$?$