compositions (version 1.40-2)

plot.acomp: Ternary diagrams

Description

Displaying compositions in ternary diagrams

Usage

# S3 method for acomp
plot(x,...,labels=names(x),
          aspanel=FALSE,id=FALSE,idlabs=NULL,idcol=2,center=FALSE,
          scale=FALSE,pca=FALSE,col.pca=par("col"),margin="acomp",
          add=FALSE,triangle=!add,col=par("col"),axes=FALSE,
          plotMissings=TRUE,
          lenMissingTck=0.05,colMissingTck="red",
          mp=~simpleMissingSubplot(c(0,1,0.95,1),
                                  missingInfo,c("NM","TM",cn)),
          robust=getOption("robust"))
 # S3 method for rcomp
plot(x,...,labels=names(x),
          aspanel=FALSE,id=FALSE,idlabs=NULL,idcol=2,center=FALSE,
          scale=FALSE,pca=FALSE,col.pca=par("col"),margin="rcomp",
          add=FALSE,triangle=!add,col=par("col"),axes=FALSE
          ,plotMissings=TRUE,
          lenMissingTck=0.05,colMissingTck="red",
          mp=~simpleMissingSubplot(c(0,1,0.95,1),
                                   missingInfo,c("NM","TM",cn)),
          robust=getOption("robust"))
 # S3 method for ccomp
plot(x,...)

Arguments

x

a dataset of a compositional class

further graphical parameters passed (see par)

margin

the type of marginalisation to be computed, when displaying the individual panels. Possible values are: "acomp", "rcomp" and any of the variable names/column numbers in the composition. If one of the columns is selected each panel displays a subcomposition given by the row part, the column part and the given part. If one of the classes is given the corresponding margin acompmargin or rcompmargin is used.

add

a logical indicating whether the information should just be added to an existing plot. If FALSE a new plot is created

triangle

a logical indicating whether the triangle should be drawn

col

the color to plot the data

labels

the names of the parts

aspanel

logical indicating that only a single panel should be drawn and not the whole plot. Internal use only

id

logical, if TRUE one can identify the points like with the identify command.

idlabs

a character vector providing the labels to be used with the identification, when id=TRUE

idcol

color of the idlabs labels

center

a logical indicating whether a the data should be centered prior to the plot. Centering is done in the choosen geometry. See scale

scale

a logical indicating whether a the data should be scaled prior to the plot. Scaling is done in the choosen geometry. See scale

pca

a logical indicating whether the first principal component should be displayed in the plot. Currently, the direction of the principal component of the displayed subcomposition is displayed as a line. In a future, the projected principal componenent of the whole dataset should be displayed.

col.pca

The color to draw the principal component.

axes

Either a logical wether to plot the axes, or numerical enumerating the axes sides to be used e.g. 1 for only plotting the lower axes, or a list of parameters to ternaryAxis.

plotMissings

logical indicating that missingness should be represented graphically. Componentes with one missing subcomponent in the plot are represented by tickmarks at the three axis. Components with two or three missing components are only represented in a special panel drawn according to the mp parameter if missings are present. Missings of type BDL (below detection limit) are always plotted, even if plotMissings is false, but in this case this fact is not specially marked. In rcomp geometry an actuall 0 in the data is never treated as missing.

lenMissingTck

length of the tick-marks to be plotted for missing values. If 0 no tickmarks are plotted. Negative lengths point outside. length 1 draws right through to the opposit corner. Missing ticks in acomp geometry are inclined showing the line of possible values in acomp geometry. Missingticks in rcomp-geometry are vertical to the axis representing the fact that only the other component is unkown. That these lines can leave the plot is one of the odd consequences of rcomp geometry.

colMissingTck

colors to draw the missing tick-marks. NULL means to take the colors specified for the observations.

mp

A formula providing a call to a function plotting informations on the missings. The call is evaluted in the environment of the panel plotting function and has access (among others) to: cn the names of the components in the current plot, x the dataset of the current plot, y the transformed dataset, (c60,s60) coordinates of the upper vertex of the triangle. missingInfo is a table giving the number of observations of the types NM=Non Missing, TM=Totally missing (i.e. at least two components of the subcomposition are missing), and the three single component missing possibilities for the three components.

robust

A robustness description. See robustnessInCompositions for details. The option is used for centering, scaling and principle components.

Details

The data is displayed in ternary diagrams. Thus, it does not work for two-part compositions. Compositions of three parts are displayed in a single ternary diagram. For compositions of more than three components, the data is arranged in a scatterplot matrix through the command pairs.

In this case, the third component in each of the panels is chosen according to setting of margin=. Possible values of margin= are: "acomp", "rcomp" and any of the variable names/column numbers in the composition. If one of the columns is selected each panel displays a subcomposition given by the row part, the column part and the given part. If one of the classes is given the corresponding margin acompmargin or rcompmargin is used.

Ternary diagrams can be read in multiple ways. Each corner of the triangle corresponds to an extreme composition containing only the part displayed in that corner. Points on the edges correspond to compositions containing only the parts in the adjacent corners. The relative amounts are displayed by the distance to the opposite corner (so-called barycentric coordinates). The individual portions of any point can be infered by drawing a line through the investigated point, and parallel to the edge opposite to the corner of the part of interest. The portion of this part is constant along the line. Thus we can read it on the sides of the ternary diagram, where the line crosses its borders. Note that these isoPortionLines remain straight under an arbitrary perturbation.

ccomp ternary diagrams are always jittered to avoid overplotting.

References

Aitchison, J. (1986) The Statistical Analysis of Compositional Data Monographs on Statistics and Applied Probability. Chapman & Hall Ltd., London (UK). 416p.

Aitchison, J, C. Barcel'o-Vidal, J.J. Egozcue, V. Pawlowsky-Glahn (2002) A consise guide to the algebraic geometric structure of the simplex, the sample space for compositional data analysis, Terra Nostra, Schriften der Alfred Wegener-Stiftung, 03/2003

Billheimer, D., P. Guttorp, W.F. and Fagan (2001) Statistical interpretation of species composition, Journal of the American Statistical Association, 96 (456), 1205-1214

Pawlowsky-Glahn, V. and J.J. Egozcue (2001) Geometric approach to statistical analysis on the simplex. SERRA 15(5), 384-398

http://ima.udg.es/Activitats/CoDaWork03

http://ima.udg.es/Activitats/CoDaWork05

See Also

plot.aplus, plot3D (for 3D plot), kingTetrahedron (for 3D-plot model export), qqnorm.acomp,boxplot.acomp

Examples

Run this code
# NOT RUN {
data(SimulatedAmounts)
plot(acomp(sa.lognormals))
plot(acomp(sa.lognormals),axes=TRUE)
plot(rcomp(sa.lognormals))
plot(rcomp(sa.lognormals5))

plot(acomp(sa.lognormals5),pca=TRUE,col.pca="red")
plot(rcomp(sa.lognormals5),pca=TRUE,col.pca="red",axes=TRUE)
# }

Run the code above in your browser using DataCamp Workspace