Information Technology Reference
In-Depth Information
3.7.5 Function
PCA.predictivities
A function for calculating the various measures of fit discussed in this chapter for
PCA biplots. It has the four arguments, discussed in Section 3.7.1 for
PCAbipl
:
X
,
X.new.samples = NULL
,
X.new.vars = NULL
,and
scaled.mat = FALSE
.Itis
assumed that
X
is an
n
×
p
data matrix with
p
not larger than
n
.Bydefault
X
is centred
but not scaled.
The various measures of fit are provided for dimensions 1, 2,
...
,
p
and the output
(value) of
PCA.predictivities
is a list with the following components.
Overall quality of the display calculated
according to (3.17).
Quality
The weights calculated according to (3.21).
Weights
Sample predictivities for original
n
samples
as defined in (3.20).
Sample.predictivities.original
Adequacies calculated according to (3.18).
Adequacies
Axes predictivities for original
p
variables as
defined in (3.19).
Axis.predictivities.original
Sample predictivities for new samples as
defined in (3.22).
Sample.predictivities.new
Axes predictivities for new variables as
defined in (3.28).
Axis.predictivities.new.1
Axes predictivities for new variables as
defined in (3.29).
Axis.predictivities.new.2
3.7.6 Function
PCA.predictions.mat
The function
PCA.predictions.mat
takes the three arguments:
X
,
scaled.mat =
FALSE
and
e.vects = 1:2
as described in Section 3.7.1. Input
scale(X)
to obtain
predictions for the
X
-values centred to zero means and unit variances;
X
and
scaled.mat
= FALSE
to obtained predictions for the unscaled
X
-values; or
X
and
scaled.mat =
TRUE
to obtain predictions for the scaled values that are transformed into the origi-
nal scales of measurement. Predictions can be obtained in any dimension by a proper
specification of argument
e.vects
.
3.7.7 Function
vector.sum.interp
A function call to
vector.sum.interp
results in interpolating a new sample into a
PCA biplot as described in Section 3.2.2. The following arguments are available:
ver-
tex.points = 3
,
p = vertex.points
,
pch.centroid = 15
,
col.centroid =
