Two-way tables: biadditive biplots - Understanding Biplots - page 266

Usage

biad.predictivities (X = wheat.data, e.vects = 1:ncol(X),
X.new.rows = NULL, X.new.columns = NULL, add.maineffect =
FALSE, biad.variant = c("Xmat", "XminMeanMat",
"InteractionMat"), predictions.dim = c("All", 1:min(nrow(X),
ncol(X))))

Arguments

Required argument. It must be in the form of a p × q
matrix representing a two-way table of size p × q being
observations (measurements) on a response variable
depending on two factors. It is assumed that p ≥ q .If
this is not the case X must be transposed.

X

Vector specifying the eigenvectors to be used for the quality
calculations. The default is to use all eigenvectors.

e.vects

If not NULL it must be a matrix of size k × q representing
k new rows interpolated into the biadditive biplot.

X.new.rows

If not NULL it must be a matrix of size p × m representing
m new columns (or axes) interpolated into the biadditive
biplot.

X.new.columns

Logical value requesting or suppressing adding of main
effects for calculated predicted values when biplotting
the interaction matrix. Defaults to FALSE.

add.maineffect

One of "Xmat" , "XminMeanMat" , "InteractionMat"
specifying whether to use the two-way table itself, the
two-way table with the overall mean subtracted or the
interaction matrix.

biad.variant

"All" or integer value requesting predictions in dimension
"integer" . Default of "All" requests predictions in all
dimensions.

predictions.dim

Value
biad.predictivities returns a list with the following named components:

Overall quality in 1, 2, ... , q dimensions. Quality

Column adequacies in 1, 2, ... , q dimensions. Column.adequacies

Column predictivities in 1, 2, ... , q dimensions.

Column.
predictivities

Row adequacies in 1, 2, ... , q dimensions. Row.adequacies

Row predictivities in 1, 2, ... , q dimensions. Row.predictivities

The column main effects.

main.effects.
columns

The row main effects.

main.effects.rows