$($a$)$ You decide to split your data into $3$ data sets:

training, validation and testing. A junior engineer

suggests that you randomly allocate data points to

each of these $3$ data and not sequentially. Would

this be a good strategy? Motivate your answer.

$($b$)$ Given the data set, a junior engineer suggests

that you use a $5^{th}$ order polynomial function as a

regression model to anticipate annual company

profit, based on the number of operational years.

Explain the risk of your model overfitting the data.

$($c$)$ Explain how overfitting can be detected given your

training, validation and test sets?

$($d$)$ Suppose you want to use a high$-$order polynomial

function to model the data set. What method can

be used to ensure that the polynomial delivers a

better fit? Explain how this method accomplishes

this and name a specific machine learning tech$-$

nique that can be applied to avoid overfitting.