Additional variables and adjusted estimates with arbitrary known variance-covariance structure

When additional variables are fitted in a linear model under arbitrary known variance-covariance struture, the extra sum of squares due to fitting the new variables and adjusted parameter estimates can be computed in an efficient manner without actually explicity fitting the entire augmented model. When the additional variables are specific dummy variables, downdating formulae are readily obtained, thus generatina methods which are well known for the linear model with variance-covariance structure σ²I. Two different methods to downdate a linear model are presented.     

The efficient cross-validation of principal components applied to principal component regression

The cross-validation of principal components is a problem that occurs in many applications of statistics. The naive approach of omitting each observation in turn and repeating the principal component calculations is computationally costly. In this paper we present an efficient approach to leave-one-out cross-validation of principal components. This approach exploits the regular nature of leave-one-out principal component eigenvalue downdating. We derive influence statistics and consider the application to principal component regression.