Lecture 5-1

Question 1

Main goal of subset selection when we talk about linear regression is to reduce RSS

Answer 1

True

Question 2

We want to maximize adjusted Rsquare not minimize it

Answer 2

True

Question 3

In subset selection we go from k=1 until p predictors and at each step we will fit the model with exactly k predicotrs , then we pick the best one

Answer 3

The best one willl give us smallest RSS or largest adjusted Rsquare

Question 4

Forward stepwise selection, we start from null model then we go from k to p-1 and at each step we fit p-k model

Answer 4

True

Question 5

Backward stepwise selection

Answer 5

We start by full model and we remove the one which is contributing the least to our model

Question 6

Shrinkage method

Answer 6

Fits model including all p and use technique to constraint coefficient estimate main goal is to reduce variance

Question 7

In shrinkage approach we have two methods

Answer 7

Ridge regression and Lasso

Question 8

In ridge regression our goal is to minimize RSS and and shrinkage term penalty (tends to be small when coefficients are close to zero)

Answer 8

True