2. Consider a linear smooth f = Sy where S is a linear smoother and y = (Y1, . .., Ym)?. Let b = f - E(Sy) be the bias vector. Show that n 1 MSE = var(f(Xi)) + - n n trace(SST) bib = 02 + n n 1 PSE = [MY* - f(Xi)}2 n i=1 trace(SS ) 1 2 bb = $1+ + n n where Y* is a new observation