Resampling, linear model selection, moving beyond linearity

A Leave-one-out cross-validation is a particular case of K-fold cross-validation

• True

• False

The predictors identified by the k step of best subset selection are the subset of the k+1 step of best subset selection

• True

• False

Which proportion of observations is on average not a part of a finite bootstrap sample

• 1/4

• 1/3

• (1-1/n)^n

• 1/e

Which model would have the smallest training residual sum of squares at the k step?

• Best subset selection

• Forward stepwise selection

• Forward stagewise selection

• Backward stepwise selection

The LASSO relative to least squares is

• More flexible and improves prediction if bias increase < variance decrease

• More flexible and improves prediction if variance increase < bias decrease

• Less flexible and improves prediction if bias increase < variance decrease

• Less flexible and improves prediction if variance increase < bias decrease

The form of the prior beliefs area of the ridge regression is

• Diamond

• Triangle

• Square

• Circle

Which of the following are some of the regularization methods?

• Ridge regression

• LASSO

• Least squares

• All mentioned

Solving the LASSO problem the training sample residual sum of squares will be

• Inverted U shaped

• U shaped

• Constant

• Steadily increasing

Which non-linear method is the most affected by the high leverage points?

• Polynomial regression

• Step functions regression

• Local regression

• Smoothing splines

Which non-linear approach does not depend on the number of knots?

• Step functions regression

• Linear splines

• Smoothing splines

• Cubic splines