The statistics of cross-validation residuals - Theory
The statistics of cross-validation residuals - Theory
The expected values of the residuals derived in the previous section
may be used to give estimated values of the generalised R-factor,
, defined by
We define 
as 
based on p excluded
reflections. If there are only random and uncorrelated errors then the
numerator of may be approximated by 
and using equation (8) we can write
We define 
as based on all f included
reflections and if the weights have been scaled so that the residual in the
numerator is equal to its expected value then
The statistics of cross-validation residuals - Theory
