Hi,
I have a question concerning the calculation of the PCAs:
I wonder where the constant factor comes from that distinguishes the
R PCAs from gretl PCAs.
<hansl>
matrix Z = {sqrt(2),sqrt(3/2);-sqrt(0.5),0;-sqrt(0.5),-sqrt(3/2)}
matrix Sig = mcov(Z)
matrix F = princomp(Z,2,1)
matrix V={}
eval eigengen(Sig,&V)
eval Z*V # the PCAs R computes with
princomp (and which I expected)
eval princomp(Z,2,1) # the PCAs gretl gives
eval Z*V/sqrt(3/2) # factor? 3=dataset obs, 2=#of vars ?
, seems like a standardization right?
<hansl>
Cheers
Leon