Information Technology Reference
In-Depth Information
Gender
Male
Hair
Education
Jeremy
England
George
Dark
Alisdair
Manual
University*Postgrad
Grey*Brown
Ivor
Harriet
Clerical*Professional
Jane
Scotland
Work
School
Wales
Female
Region
Myfanwy
Fair
Figure 8.13
Categorical PCA. The variables
Hair
,
Education
and
Work
are treated as
ordinal and the variables
Gender
and
Region
are treated as nominal. The axes for the
nominal variables are colour-calibrated as in Figure 8.11. Axes for the ordinal variables
are calibrated in the same colour (here green) but with thickness increasing with order.
Note the way ties are written, e.g.
University
*
Postgrad, Clerical
*
Professional
.
Arguments with a specific meaning in
MCAbipl
Required argument. An
n
p
data matrix where each
column represents a categorical variable, e.g.
Table 8.1.
×
X
One of “indicator”, “Burt”, “EMC”. Defaults to
“indicator”.
mca.variant
The titles to be used for each set of prediction regions.
Default is to use the column names of
X
.
main
If TRUE, the prediction regions associated with each
categorical variable are constructed in separate graph
windows. The actual construction of the prediction
regions is performed by the functions
pred.regions
and
Colour.Pixel
that are called by
MCAbipl
.See
also Section 4.7.1. Default is FALSE.
prediction.regions
Symbol used in pixel colouring. Default is 15 (solid
square).
pred.region.symbol
