PCA.plot: Principal Component Analysis plotting functions

Description

Plotting functions for PCA: for scores, loadings, scores and loadings simultaneously (a biplot), and variances (a screeplot, where the log of the explained variance is plotted for each PC).

Usage

# S3 method for PCA
scoreplot(object, pc = c(1, 2), pcscores = scores(object),
          show.names = FALSE, xlab, ylab, xlim, ylim, …)
# S3 method for PCA
loadingplot(object, pc = c(1, 2), pcloadings = loadings(object),
            scalefactor = 1, add = FALSE, show.names = FALSE,
            xlab, ylab, xlim, ylim, col = "blue", min.length =
            0.01, varnames = NULL, …)
# S3 method for PCA
biplot(x, pc = c(1,2),
       show.names = c("none", "scores", "loadings", "both"),
       score.col = 1, loading.col = "blue",
       min.length = .01, varnames = NULL, …)
screeplot(object, type = c("scree", "percentage"), npc, …)

Arguments

object, x

an object of class "PCA" (see below).

which PCs to show.

pcscores

matrix of scores, by default the scores of the PCA model object.

show.names

show names rather than plotting symbols. For loadingplot and scoreplot a logical (default: FALSE), for biplot one of 'scores', 'loadings', 'both' or 'none' (default).

xlab, ylab, xlim, ylim, col

graphical parameters of the plot.

pcloadings

matrix of loadings, by default the loadings of the PCA model object.

scalefactor

scaling factor for the loadings; used internally, when the loadingplot function is called from within biplot.PCA.

add

logical, whether to add to the existing plot (again, useful when loadingplot is called from within biplot.PCA).

npc

how many PCs to show in the scree plot (starting from 1).

type

show a real screeplot (scree) or show the percentage of variance explained (percentage).

score.col, loading.col

colours of the scores and loadings in a biplot.

min.length

minimal length of loading vectors to be plotted by arrows. Vectors that are too short lead to warning messages, are not interesting, and only clutter the graphic.

varnames

alternative vector of variable names.

…

Graphical arguments passed on to lower-level plotting functions.

Details

Score plots and loading plots show the amount of explained variance at the axis labels only when PCA has been performed at mean-centered data.

References

R. Wehrens. "Chemometrics with R - Multivariate Data Analysis in the Natural Sciences and Life Sciences". Springer, Heidelberg, 2011.

Examples

Run this code

# NOT RUN {
data(wines, package = "ChemometricsWithRData")
wines.PC <- PCA(scale(wines))
scoreplot(wines.PC, col = wine.classes, pch = wine.classes)
loadingplot(wines.PC, show.names = TRUE)
biplot(wines.PC, score.col = wine.classes)
screeplot(wines.PC)
# }

Run the code above in your browser using DataLab

Last chance! 50% off unlimited learning