Let Y=[Y1, ..., Y_v] be a list of variables, of which
n observations are known. Suppose the observations are
standardize so that
mean(Y_j)=0
var(Y_j) = 1
for j=1, .., v. The principal components P=[P1,.., P_v] relate to
Y by:
Y = P L
where L is the "loadings" matrix. We can get P and L easily
with proc princomp.
Suppose one does some modeling work on the P coordinates.
Is there a procedure that will calculate the value Y' that
corresponds to a fitted value of P' ?
VZ
.
.
=================================================================
Instructions for joining and leaving this list, remarks about the
problem of INAPPROPRIATE MESSAGES, and archives are available at:
. http://jse.stat.ncsu.edu/ .
=================================================================
