biplot.choice: Biplot choice tool

In tpepler/cpc: Common principal component (CPC) analysis and applications

Biplot choice tool

Description

Provides quality measures for different types of principal component biplots of grouped data, to select the best method for constructing a principal component biplot for any specific data set.

Usage

biplot.choice(datalist, rdim, add.projectmats = NULL)

Arguments

datalist	List of the data for which the biplot is to be constructed, created with a command such as list(group1, group2, ...).
rdim	Number of dimensions of the biplot representation.
add.projectmats	(Optional) Additional orthogonal projection matrices for which to compute the biplot quality measures [NOT YET IMPLEMENTED].

Details

This function calculates a number of quality measures relevant in judging the appropriateness of a principal component biplot for grouped data. The user would then choose the method leading to the best biplot for the purpose of interest.

Quality measures are automatically calculated for biplots constructed from the following orthogonal projection matrices:

'Pooled S' – eigenvectors of the pooled data (groups uncentred)

'Pooled data' – eigenvectors obtained from the pooled covariance matrix

'Flury' – common eigenvector matrix estimated with the FG algorithm

'Stepwise CPC' – common eigenvector matrix estimated with the stepwise CPC algorithm

'JADE' – common eigenvector matrix estimated with the rjd function in the JADE package

Value

Returns a matrix of results with the columns:

Overall	Overall quality of the biplot display.
Within	Mean quality of the within-group variation as displayed in the biplot.
Between	Quality of the between-group variation as displayed in the biplot.
Adequacy	Median adequacy of the variables.
MSPE	Mean of the mean standard predictive errors (MSPE) of the variables.
Sample predictivities	Mean predictivity of the observations as displayed in the biplot.

Author(s)

Theo Pepler

References

Pepler, P.T. (2014). The identification and application of common principal components. PhD dissertation in the Department of Statistics and Actuarial Science, Stellenbosch University.

See Also

See biplot.measures for more details on the different quality measures for principal component biplots of grouped data.

Examples

# The Iris data
data(iris)
setosa <- iris[1:50, 1:4]
versicolor <- iris[51:100, 1:4]
virginica <- iris[101:150, 1:4]

# For a 2-dimensional biplot
biplot.choice(datalist = list(setosa, versicolor, virginica), rdim = 2)

# For a 3-dimensional biplot
biplot.choice(datalist = list(setosa, versicolor, virginica), rdim = 3)

tpepler/cpc documentation built on July 7, 2022, 2:13 a.m.