1. Assume the training feature matrix X is a (p+1) N matrix, where p is the number of predictors and N is the sample size of training data, and p>N. a) Assume X is of full rank, i.e. rank(X)=r=N