Information Technology Reference
In-Depth Information
The distance function for calculating that part of
the
D
matrix contributed by the quantitative
variables. In the current implementation, one of
“
Pythagoras
”, “
Clark
”or“
SqrtL1
”. Can
easily be extended to any qualifying distance
metric as described for
Nonlinbipl
. Default is
“
Pythagoras
”.
dist.cont
The distance function for calculating that part of
the
D
matrix contributed by the qualitative
variables. In the current implementation, EMC.
dist.cat
Plotting symbol used for colouring pixels when
prediction regions are constructed.
plot.symbol
Size of
plot.symbol
.
plot.symbol.size
Logical TRUE or FALSE requesting or
suppressing construction of prediction regions.
prediction.regions
One of “
circle
”, “
normal
”or“
back
”. Default is
“
normal
”.
prediction.type
If set to TRUE when
dist
=
“
Pythagoras
”
sample predictions are obtained as with
PCAbipl
. Should be set to FALSE for all other
distance metrics.
straight
Numerical value for specifying the horizontal grid
size for constructing prediction regions.
x.grid
Numerical value for specifying the vertical grid
size for constructing prediction regions.
y.grid
Vector for specifying a colour for each category of
a qualitative variable.
colours.pred.regions
Details
In principle
Genbipl
combines the capabilities of
Nonlinbipl
and
MCAbipl
. It calls the
functions
drawbipl.genbipl
and
pred.regions
for actually constructing the plots.
Note that the distance from a pixel to its nearest CLP is calculated in the full space.
Value
In addition to the construction of a generalized biplot and its associated prediction regions,
Genbipl
outputs a list with the following components:
A matrix containing the coordinates of the samples
in the biplot space.
Z
A list consisting of
p
1
matrices - one matrix for
each quantitative variable containing the details
for constructing its biplot axis.
Z.axes
A vector containing all the eigenvalues of the PCO
of the
ddistance
matrix (9.6).
e.vals
The
ddistance
matrix
D
(
1
)
in (9.6).
D.cont
