Skip to content

Dot plot similar to StatKey

6 messages · David Arnold, Jim Lemon, S Ellison +2 more

Original
David Arnold
# 
Hi,

I'd like to use R to produce the following plot:

dotplot.jpeg <http://r.789695.n4.nabble.com/file/n4672628/dotplot.jpeg>  

This was constructed using StatKey at:

http://www.lock5stat.com/statkey/bootstrap_1_quant/bootstrap_1_quant.html
<http://www.lock5stat.com/statkey/bootstrap_1_quant/bootstrap_1_quant.html>  

The data used was AtlantaCommute (Time) and I generated 1000 bootstrap
samples of size 500 with replacement. I'd really like to learn how to make
such a dotplot in R.

Any thoughts?

Thanks.



--
View this message in context: http://r.789695.n4.nabble.com/Dot-plot-similar-to-StatKey-tp4672628.html
Sent from the R help mailing list archive at Nabble.com.
Jim Lemon
#
On 07/30/2013 04:21 PM, David Arnold wrote:
Hi David,
Have a look at the dotplot.mtb function in plotrix.

Jim
S Ellison
# 
x<-rnorm(500)
xr <- round(x, 1)
stripchart(xr, method="stack", pch=19) 

will do this if the data are rounded appropriately. You may have some fun with 'appropriately'.

Tinker with ylim to get the axis limits right; stripchart's baseline is at 1 for a single group.

S Ellison
 



*******************************************************************
This email and any attachments are confidential. Any use...{{dropped:8}}
John Kane
# 
geom_dotplot() in the ggplot2 package perhaps?
ggplot(mtcars, aes(x = mpg)) + geom_dotplot()

John Kane
Kingston ON Canada
____________________________________________________________
GET FREE SMILEYS FOR YOUR IM & EMAIL - Learn more at http://www.inbox.com/smileys
Works with AIM?, MSN? Messenger, Yahoo!? Messenger, ICQ?, Google Talk? and most webmails
S Ellison
# 
I said:
A bit of tinkering gives something that gets close to 'appropriate' (below); you may want to tinker with the tuning factor (or, of course, write something entirely different).

#
# Based on ?points: For 'pch' in '0:25' the default size is about
#          75% of the character height (see 'par("cin")')

round.pch<-function(x, tune=1.0) {
	#produces data rounded to approximate graphics symbol width
                is.window <- !is.null(dev.list()[1])
	if(is.window) {
		psize.ins <- 0.75 * par("cin")[2]/(2 * tune)
		psize.usr <- psize.ins*diff(par("usr")[1:2] / par("pin")[1])
	} else {
		psize.usr <- diff(range(pretty(x))) * 0.15 / (14 * tune)
		               #0.15 is the default par("cin")[2] * 0.75
	}
	return( psize.usr * (floor(x/psize.usr) + 0.5) ) 
}

set.seed(73)
x<-rnorm(500)
stripchart(round.pch(x), method="stack", pch=19, ylim=c(0.9,2))


round.pch above uses a vaguely reasonable default rounding with no graphics device present and will use the present device size and units if you have already called plot.new and set user coordinates. Odd things wll happen with a device open and no user coordinates set. Calling this repeatedly with no xlim set in stripchart will cause results to vary visibly as the user coordinates change slightly for each call.

*******************************************************************
This email and any attachments are confidential. Any use...{{dropped:8}}
Richard M. Heiberger
# 
I would use panel.dotplot.tb from the HH package.
It is based on lattice, hence will plot multiple groups on the same scale.

continuing with S Ellison's example

## install.packages("HH") ## if necessary
library(HH)
rpx <- round.pch(x)
dotplot( ~ rpx, panel=panel.dotplot.tb, ylim=c(.98, 1.15))
dotplot( ~ rpx | factor(rep(1:2, each=250)), panel=panel.dotplot.tb,
layout=c(1,2), ylim=c(.98, 1.15))

Rich
On 7/30/13, S Ellison <S.Ellison at lgcgroup.com> wrote: