plot.groc: Plot groc objects.

In groc: Generalized Regression on Orthogonal Components

Description

A function to plot groc objects.

Usage

## S3 method for class 'groc'
plot(x, h=x$ncomp, cex=0.8, ...)

Arguments

x	A groc object.
h	Number of components in the model.
cex	Character expansion factor for point labels.
...	Further arguments passed to internal plot function.

Details

If plsrob=FALSE, a plot of robust Mahalanobis distances for residuals versus robust Mahalanobis distances for components. Useful for identification of good points, vertical outliers, good and bad leverage points.

If plsrob=TRUE, the previous plot is done with another similar plot of classical Mahalanobis distances to compare the identification of the various type of points obtained by classical or robust partial least squares.

Author(s)

Martin Bilodeau ([e-mail]) and Pierre Lafaye de Micheaux ([e-mail])

References

Martin Bilodeau, Pierre Lafaye de Micheaux, Smail Mahdi (2015), The R Package groc for Generalized Regression on Orthogonal Components, Journal of Statistical Software, 65(1), 1-29,
https://www.jstatsoft.org/v65/i01/

Examples

## This example takes some time:
## Not run: 
data("pulpfiber",package="robustbase")
X <- as.matrix(pulpfiber[, 1:4])
Y <- as.matrix(pulpfiber[, 5:8])
data <- data.frame(X=I(X), Y=I(Y))
set.seed(55481)
out.rob <- groc(Y ~ X, data=data, plsrob=TRUE)
plot(out.rob, cex=.6)

## End(Not run)

groc documentation built on Dec. 18, 2020, 9:07 a.m.