In Section 4.7.3, we consider regressing y on a set of principal components, rather than the original data. For simplicity, assume that X does not contain a constant term, and that the K variables are measured in deviations from the means and are “standardized” by dividing by the respective standard deviations. We consider regression of y on L principal components, Z = XC_L, where L < K. Let d denote the coefficient vector. The regression model is y = Xβ + ε. In the discussion, it is claimed that E[d] = C'_Lβ. Prove the claim.