Information Technology Reference
In-Depth Information
biplot. In addition to the biplot, the call returns a matrix with the three-dimensional
predictions of all samples on all the variables. As with all other biplots, the
e.vects
argument allows for any three principal axes to be chosen for the biplot scaffolding.
3.8.6 Changing the scaffolding axes in conventional
two-dimensional PCA biplots
In Figure 3.41 we show for the copper froth data two-dimensional PCA biplots using various
principal components to form the scaffolding for the biplot. The two scaffolding axes are
selected according to the first two components of the vector assigned to argument
e.vects
of
PCAbipl
. Some structure is seen in these biplots. Note also that the predictivities of all
Scaffolding axes: 1st and 2nd
p
rincipal components
Scaffolding axes: 1st and
3
rd principal components
Scaffolding axes: 2nd and 3rd principal components
Scaffolding axes: 1st and 4th principal components
Figure 3.41
Two-dimensional PCA biplots of the copper froth data obtained for various
combinations of the two scaffolding axes. Quality of the displays: (top left) 85% (max-
imum for a two-dimensional biplot); (top right) 70.85%; (bottom left) 32.61%; (bottom
right) 64.27%.
