CHAPTER 21 . (hemometries for Sampling and Analysis: Theory and Environmental Applications
393
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
z
,
,
,
,
,
a
,/ L.!.:::!.:..:..:j!:.L,;.,;.._.:...:....::....o:......:.... _ _ _ ,", o:.....:._--x ..Y
,
,
,
'
,
/
,
,
,/
/'
,
5 13
x
20
74
z
b
E.igenvector 1
,
,
,
5
c
""'"
" '\.
"',, 10
17 ",
9 13 \,1
18
' : ~: 5 ' : """"""'"",::
Varivector 1
Fig. 21.5. a The space of original information; b The space of structured information, of PCs; c The space
of rotated useful information
vectors define this plane, and they can be used to visualise the inner space, the relevant information (Fig. 21.5b).
Note that the three original variables are projected on the same plot according their
loadings. Variable X contributes to component El (large positive loading of El on X),
whereas its contribution to component E2 is negligible. On the contrary, Y contributes
with large positive loading to component E2. Z contributes with positive loading to