[rkward-users] [Help] RE: How do I create matrix error bars?

[SourceForge.net]

The following forum message was posted by kapatp at http://sourceforge.net/projects/rkward/forums/forum/165574/topic/3704541:

Ok, here is a working example. Replace x by your dataset name.

[code]x <- data.frame (id=1:10, X1=runif(10), X2=2*runif(10), X3=5*runif(10), X4=10*runif(10), X5=20*runif(10), X6=50*runif(10))
U <- reshape(x, varying=2:7, v.names=\'moles\', direction=\'long\')
U$time <- factor(U$time); levels(U$time) <- c(1,2,5,10,20,50);
library(lattice)

# aggregate() in R >= 2.11.0 has a method for formula objects
if (getRversion() <= \"2.10.1\") {
	S <- data.frame (moles=apply(x[,2:7],2,median), time=1:6)
} else S <- aggregate (moles~time, data=U, median)

trellis.par.set(box.rectangle = list(lty = 0))
bwplot(moles~time, data=U, horizontal=FALSE, xlab=\'time\',
	stats = function (...) {
		bstats <- boxplot.stats (...)
		bstats$stats[c(2,4)] <- bstats$stats[3]
		return(bstats)
	},
	panel = function (x,y,...) {
		panel.bwplot (x,y,...)
		panel.xyplot(x=S$time, y=S$moles,type=\'l\')
	}
)[/code]

If this much of coding puts you off then [b]R[/b] is not the right software for you!

Now, here is a bit of illuminating rant:
As Thomas pointed out earlier,  Rkward is merely a GUI; [b]R[/b] is the main statistical engine behind [b]Rkward[/b]. The minimally available menu driven (statistical) functions in Rkward is ofcourse driven by appropriate R functions/code underneath. [b]Again, Rkward is not a statistical software.[/b] So, if you are used to menu driven softwares, such as, SPSS, SigmaPlot (as you mentioned), Minitab, StatXact, STATISTICA (even SAS and Splus to a good extent), then I am sorry to spoil your surprise, [b]with R[/b] you are in for a rude shock (of course, depends on how much of data handling you do): 
* R is completely comandline driven. 
* You could use R from a just a terminal (xterm, aterm, uterm, konsole, gnome-terminal, ...). 
* If you have used the emacs editor then you could run R (or Splus) from withing emacs (using ESS)
* You could use kate (the KDE editor) and link with an R session w/o using RKward (it is a pain though)
* There is a similar plugin for gedit as well..
* http://en.wikipedia.org/wiki/R_%28programming_language%29

To repeat: RKward will not hold your hand through your statistical analysis! As you can see from the code above, it has nothing to do with RKward!!

Good luck.