adaptiveGPCA (version 0.1.3)

gpcaFullFamily: Make a sequence of ordinations

Description

Creates a sequence of gPCA data representations. One end of the sequence (\(r = 0\)) doesn't do any regularization according to the variable structure (and so is just standard PCA), and the other (\(r = 1\)) does a maximal amount of regularization according to the variable structure.

Usage

gpcaFullFamily(X, Q, weights = rep(1, nrow(X)), k = 2, rvec = (0:100)/100,
  findReflections = TRUE, returnLong = FALSE, sampledata = NULL,
  variabledata = NULL)

Value

A list containing elements for the sample points (locationList), the species points (speciesList), and the variance fractions (varsList). Each element is itself a list of data frames (location/species points) or of vectors (for the variances).

Arguments

X

A data matrix of size \(n \times p\).

Q

A \(p \times p\) similarity matrix defining an inner product on the rows of X.

weights

A vector of weights for the rows of X.

k

The number of components to compute for each ordination.

rvec

The values of \(r\) for which to make the ordinations.

findReflections

Whether or not flip the axes so as to make neighboring ordinations as close as possible. If k is very large this should be false since all possible axis combinations are searched over.

returnLong

Return a long data frame with the samples/variables instead of a list of data frames.

sampledata

Extra sample data to be included along with the sample scores.

variabledata

Extra variable data to be included along with the variable loadings.

Examples

data(AntibioticSmall)
out.ff = gpcaFullFamily(AntibioticSmall$X, AntibioticSmall$Q, k = 2)