Answer the following statements with TRUE or FALSE.

?

(16) In linear regression, BIC works the same way as AIC, but in general selects a smaller model than AIC does. (17) LASSO can only work when the number of covariates (or dimension) p is larger than the sample size n. (18) Let X be an x p matrix with n < p. The inverse of XTX does NOT exist. (19) The inverse (X"X+AI)-l in the ridge estimator 3,idge always exists for any A> 0. (X"X+AI)-'X"Y (20) With appropriate tuning, LASSO estimators can have zero values in the esti- mated regression coefficients.