Information Technology Reference
In-Depth Information
By default a list is returned. Each element of the list is a
matrix of the same size as the input two-way table
containing the specified predicted values. The first matrix
contains the predictions in the first dimension, the second
one in the first two dimensions, and so on.
predictions.rows
Predictivities in 1, 2,
,
q
dimensions for each of the
k
rows given in the matrix
X.new.rows
.
...
New.Row.
Predictivities
Predictivities in 1, 2,
,
q
dimensions for each of the
m
columns given in the matrix
X.new.columns
.
...
New.Column.
Predictivities
6.6.3 Function
biad.ss
Description
This function is for calculating the partitioning of the total sum of squares associated
with a two-way table. This function is intended for use together with
biadbipl
.Itis
called with one argument, the input
X
matrix to
biadbipl
.
Value
The function
biad.ss
returns a list with the following named components:
Input
X
.
X
The estimated (approximated)
X
in the two-dimensional
biplot space. Main effects are estimated according to
(6.3), while the interactions defined by (6.4) and
assembled into the matrix
Z
are approximated by the
singular vector pairs associated with the two largest
singular values of
Z
.
X.hat
Interaction matrix.
Interaction.mat
1
/
2
where the interaction matrix is
U
V
.
Matrix
U
C.mat
1
/
2
where the interaction matrix is
U
V
.
Matrix
V
D.mat
Total sum of squares for interaction.
SS.Interaction.
Total
The break-down of
SS.Interaction.Total
into two
degrees of freedom components.
SS.Interaction.
components
Cumulative sum of the
SS.Interaction.components
.
cumsum
Complete ANOVA table with components as in Table 6.3.
ANOVA.Table
6.7 Examples of biadditive biplots: the wheat data
We begin by presenting a biplot of Table 6.1. This biplot (see Figure 6.2), approximating
the overall mean, both main effects and interaction part of the model, is not very exciting
