D Question 31 4 pts Consider the special case with ou = 0 but of # 0. Given X, you are asked to predict Y using a linear function of X. What is the best linear function that minimizes the MSE? OX O X/2 Not enough information is given It is neither X nor X/2 D Question 32 4 pts Consider the special case with of = 0 but ou # 0. Given X, you are asked to predict Y using a linear function of X. What would be the best linear function that minimizes the MSE? It is neither X nor X/2 Ox Not enough information is given O X/2D Consider the following causal model y + 3 - U; a + 2y - v. Suppose the values of (u, v) are generated from correlated normal distributions so that U ~ N(0, ou) V ~ N (0 , (7 ) cou( U, V) = Ju for some constants ou, of and Ouv. Given the above inputs (U, V), the random variables (X, Y) are the output of the above causal system. That is, (X, Y) solves Y + X -U; X + 2Y - V. That is, X = 2U + V; Y - U+V. While (X, Y) are potentially observable, (U, V) are not