[R] 2 Biplot Questions
Wayne Aldo Gavioli
wgavioli at fas.harvard.edu
Thu Jul 19 03:26:58 CEST 2007
I have 2 questions about the Biplot function:
Both of the questions refer to the following type of graph, which is a biplot of
a principal component analysis:
1. Does anyone know how to change the appearance of data points on the biplot?
As it is currently, for this type of graph, the label for each data point shows
up on the graph - does anyone know how to get rid of those labels, say, and just
make each piece of data represented by a point, the old-fashioned way? (Also, I
want to be able to keep the labels for the red arrows on the graph, just not the
data points - any suggestions? Is this even possible?)
2. What is the effect of scaling the biplot - for example, if I did:
biplot(prcomp(USArrests), scale = TRUE)
I get a different graph than if I enter the command above. I looked in the help
pages, and I see that it says that when you scale it, "the observations are
scaled by lamda ^ (1-scale) where lambda are the singular values as computed by
princomp". Can anyone explain what that means in layman's terms? Why would you
want to scale the observations by lamda - what does that accomplish? And what
is being scaled anyway: the principal components or the underlying data points?
If anyone could help, that would be great.
More information about the R-help