compositions: ratioLoadings – R documentation

Become an expert in R — Interactive courses, Cheat Sheets, certificates and more!

ratioLoadings

Loadings of relations of two amounts

Description

In a compositional dataset the relation of two objects can be interpreted safer than a single amount. These functions compute, display and plot the corresponding pair-information for the various principal component analysis results.

Usage

relativeLoadings(x,...)
## S3 method for class 'princomp.acomp'
relativeLoadings(x,...,log=FALSE,scale.sdev=NA,
                                                  cutoff=0.1)
## S3 method for class 'princomp.aplus'
relativeLoadings(x,...,log=FALSE,scale.sdev=NA,
                                                  cutoff=0.1)
## S3 method for class 'princomp.rcomp'
relativeLoadings(x,...,scale.sdev=NA,
                                                  cutoff=0.1)
## S3 method for class 'princomp.rplus'
relativeLoadings(x,...,scale.sdev=NA,
                                                  cutoff=0.1)
## S3 method for class 'relativeLoadings.princomp.acomp'
print(x,...,cutoff=attr(x,"cutoff"),
                                                  digits=2)
## S3 method for class 'relativeLoadings.princomp.aplus'
print(x,...,cutoff=attr(x,"cutoff"),
                                                 digits=2)
## S3 method for class 'relativeLoadings.princomp.rcomp'
print(x,...,cutoff=attr(x,"cutoff"),
                                                 digits=2)
## S3 method for class 'relativeLoadings.princomp.rplus'
print(x,...,cutoff=attr(x,"cutoff"),
                                                 digits=2)
## S3 method for class 'relativeLoadings.princomp.acomp'
plot(x,...)
## S3 method for class 'relativeLoadings.princomp.aplus'
plot(x,...)
## S3 method for class 'relativeLoadings.princomp.rcomp'
plot(x,...)
## S3 method for class 'relativeLoadings.princomp.rplus'
plot(x,...)

Arguments

`x`	a result from an amount PCA `princomp.acomp`/`princomp.aplus`/`princomp.rcomp`/`princomp.rplus`
`log`	a logical indicating to use log-ratios instead of ratios
`scale.sdev`	if not `NA`, a number specifying the multiple of a standard deviation, used to scale the components
`cutoff`	a single number. Changes under that (log)-cutoff are not displayed
`digits`	the number of digits to be displayed
`...`	further parameters to internally-called functions

Details

The relative loadings of components allow a direct interpretation of the effects of principal components. For acomp/aplus classes the relation is induced by a ratio, which can optionally be log-transformed. For the rcomp/rplus-classes the relation is induced by a difference, which is meaningless when the units are different.

Value

The value is a matrix of type "relativeLoadings.princomp.*", containing the ratios in the compositions represented by the loadings (optionally scaled by the standard deviation of the components and scale.sdev).

Author(s)

K.Gerald v.d. Boogaart http://www.stat.boogaart.de

Examples

data(SimulatedAmounts)
pc <- princomp(acomp(sa.lognormals5))
pc
summary(pc)
relativeLoadings(pc,log=TRUE)
relativeLoadings(pc)
relativeLoadings(pc,scale.sdev=1)
relativeLoadings(pc,scale.sdev=2)

plot(relativeLoadings(pc,log=TRUE))
plot(relativeLoadings(pc))
plot(relativeLoadings(pc,scale.sdev=1))
plot(relativeLoadings(pc,scale.sdev=2))

compositions

Compositional Data Analysis

v2.0-1

GPL (>= 2)

Authors

K. Gerald van den Boogaart <boogaart@hzdr.de>, Raimon Tolosana-Delgado, Matevz Bren

Initial release

2021-01-08