Chapter 2 (Statistical Learning & Assessing Models)

Question 1

What are the advantages of a very flexible (versus a less flexible) approach for regression or classification?

Answer 1

Obtaining a better fit for non-linear models, decreasing the bias

Question 2

What are the advantages of a very flexible (versus a less flexible) approach for regression or classification?

Answer 2

Requires estimating a greater number of parameters, follow the noise too closely (overfitting, increasing variance)

Question 3

When would a more flexible approach be preferred for regression or classification?

Answer 3

When we are interested in prediction & not interpretability

Question 4

When would a less flexible approach be preferred for regression or classification?

Answer 4

When we are interested in inference & the interpretability of the results

Question 5

What is a parametric statistical learning approach?

Answer 5

assume a form for f, so it reduces the problem of estimation a set of parameters

Question 6

What is a non-parametric statistical learning approach?

Answer 6

doesn’t assume a functional form for f & so requires a very large number of observations to accurately estimate f