(24) Perform variable selection for modell using lasso penalization. Use the glmnet() R commands and set the type of measure to be the deviance and the number of folds to be 10 for the obtaining the penalty constant A using cross-validation. For this problem, do not force the main effects to be in the model. Which of the following are correct regarding this method? a. The penalty constant A value resulting in the lowest cross-validation error was between -1.5 and -0.5. b. The first variable entering the model was A10.14. c. When log lambda is 0, fewer than 10 predicting variables are selected. d. glmnet has estimated a lower coefficient value for A5.9 than for A10.14 for all log lambda below 0. Sol: d. Solutions can be found from plotting the coefficient path. Labels correspond to the columns of X