FRB (version 2.0-1)

plot.FRBpca: Plot Method for Objects of class 'FRBpca'

Description

Plot functions for FRBpca objects: plots PC variances, PC angles and PC loadings, with bootstrap inference

Usage

# S3 method for FRBpca
plot(x, which = 1:3, pcs.loadings = 1:min(5, length(x$eigval)),
confmethod = c("BCA","basic"), ...)

plotFRBvars(x, cumul = 2, confmethod = c("BCA","basic"), 
            npcs = min(10, length(x$eigval)))
plotFRBangles(x, pcs = 1:min(12,length(x$eigval)))
plotFRBloadings(x, confmethod = c("BCA","basic"), 
            pcs = 1:min(5, length(x$eigval)), nvars=min(10, length(x$eigval)))

Value

Returns invisibly the first argument.

Arguments

x

an R object of class FRBpca, typically created by FRBpcaS or FRBpcaMM

which

integer number(s) between 1 and 3 to specify which plot is desired (1 = variances; 2 = angles; 3 = loadings)

pcs.loadings

integer number(s) indicating for which of the PCs the loadings should be shown (in case the which argument contains 2)

cumul

integer between 0 and 2: 0 = screeplot, i.e. the variances of the PCs are shown; 1 = the cumulative variances (percentage) of the PCs are shown; 2 = (default) both plots are shown on the same page

confmethod

which kind of bootstrap confidence intervals to be displayed: 'BCA'= bias corrected and accelerated method, 'basic'= basic bootstrap method

npcs

number of PCs to be included in screeplot/cumulative variances plot

pcs

PCs to consider in plot; defaults to first 12 (maximally) for plotFRBangles; defaults to first 5 for plotFRBloading (each PC is on a separate page here)

nvars

number of variables for which loadings should be shown in each PC; the loadings are shown in decreasing order in each PC

...

potentially more arguments

Author

Gert Willems and Ella Roelant

Details

The generic plot function calls plotFRBvars, plotFRBangles and plotFRBloadings, according to which of these are respectively specified in argument which, and displays the plots on separate pages (the user is prompted for each new page). The PCs for which the loadings should be plotted can be specified through the pcs.loadings argument. The other arguments are set to their default values by plot.

The solid curves displayed by plotFRBvars indicate the actual estimates of the variances (or percentages), while the dashed curves represent the confidence limits as computed by FRBpcaS or FRBpcaMM.

plotFRBangles plots, for each PC, histograms of the angles between the bootstrapped PC and the original PC estimate. The angles are in radians, between 0 and pi/2. These limits are indicated by the red vertical lines. Angles close to zero correspond to bootstrapped PCs closely aligned with the original PC, while an angle close to pi/2 means the bootstrapped PC is roughly perpendicular to the original estimate (hence a large number of angles close to pi/2 implies high variability). If the number of PCs specified in pcs is very large (usually larger than the default settings), the histograms may not fit on one page and a selection will be made (the user will be given a warning in that case).

In plotFRBloadings, the red dots represent the loadings, which are between -1 and 1. The square brackets indicate the confidence limits as computed by FRBpcaS or FRBpcaMM. Only the loadings of the first nvars variables are shown, where the variables were ordered according to the absolute value of the loading (i.e. only the nvars most important variables for that particular PC are shown).

References

• S. Van Aelst and G. Willems (2013), Fast and robust bootstrap for multivariate inference: The R package FRB. Journal of Statistical Software, 53(3), 1--32. tools:::Rd_expr_doi("10.18637/jss.v053.i03").

See Also

FRBpcaS, FRBpcaMM, summary.FRBpca

Examples

# \donttest{
    data(ForgedBankNotes)
    
    MMpcares <- FRBpcaMM(ForgedBankNotes, R=999, conf=0.95)
    plot(MMpcares) 
    
    ## a closer look at the screeplot, specifying basic bootstrap intervals
    plotFRBvars(MMpcares, cumul=0, confmethod="basic")
    
    ## plots the bootstrap angles for the first PC only
    plotFRBangles(MMpcares, pcs=1)
    
    ## plots the loadings, with basic bootstrap intervals, for *all* the PCs 
    plotFRBloadings(MMpcares, confmethod="basic", pcs=1:ncol(ForgedBankNotes))
# }