R package to automatically produce combination plot?

From: Paul Murrell <p.murrell at auckland.ac.nz>
Subject: Re: [R] R package to automatically produce combination plot?
To: jasonkrupert at yahoo.com
Cc: R-help at r-project.org
Date: Tuesday, March 17, 2009, 3:33 PM
Hi


Jason Rupert wrote:

Paul,

This is great!  Like you said it is really close.  I

made a few

changes and for some reason the y-axis label magically

came back.  I

tried to remove it but it wouldn't stay away.

You have removed the 'axes=FALSE' from the qqplot()
call.

Also, you have removed some of the explicit setting of
'ylim', 'xlim',
and 'breaks', which means that your plots are using
subtly different
coordinate systems.

Also, for some reason the title exceeds the margins of

the layout.  I

am going ot mess around with this a bit more, but any

suggestions for

fixing this are also greatly appreciated.

The title fits (just) for me.  If you need more room, bump
up the value
in the par(oma) setting.  You could also reduce the
'line' value in your
mtext() call.

Paul

Thank you again.


y<-rnorm(1000)

yrange <- range(y)

#histdata <- hist(y, breaks=breaks, plot=FALSE)

histdata <- hist(y,

plot=FALSE) densitydata <- density(y)

par(oma=rep(3, 4), mar=rep(0, 4))

layout(matrix(1:3, ncol=3), widths=c(.2, .1, 1),

respect=TRUE)

barplot(histdata$density, horiz=TRUE, space=0,

xlim=c(0,

max(histdata$density)*1.1), axes=FALSE) par(new=TRUE) 
plot(densitydata$y, densitydata$x, xlim=c(0,
max(histdata$density)*1.1), ylim=yrange,

type="l", ann=FALSE,

axes=FALSE) axis(2) box()

boxplot(y, ylim=yrange, axes=FALSE) box()

# qqnorm(y, axes=FALSE, ylim=yrange, xlim=yrange,

main="") qqnorm(y,

, main="") axis(2, labels = FALSE)

#abline(0, 1) qqline(y, col="red",

lwd=3) box()

kurtosis_val <-(sum((y - mean(y))^4

))/(var(y)*var(y))/length(y)-3

skewness<-function(x) { m_skew=mean(x)

me_skew=median(x) 

s_skew=sqrt(var(x)) sk_skew=(m_skew-me_skew)/s_skew

return(sk_skew) }

title_text<-paste("Title Text",

"\n Mean = ", format(mean(y),

digits=4, scientific=F), " Standard Deviation =

", format(sd(y),

digits=4, scientific=F), "\n Skewness =

", format(skewness(y),

digits=4, scientific=F), "Kurtosis =",

format(kurtosis_val, digits=4,

scientific=F), sep="")

mtext(title_text, NORTH<-3, line=0, adj=0.5,

cex=1.2, col="red",

outer=TRUE)






--- On Tue, 3/17/09, Paul Murrell

<p.murrell at auckland.ac.nz> wrote:

From: Paul Murrell

<p.murrell at auckland.ac.nz> Subject: Re: [R] R

package to automatically produce combination plot?

To:

jasonkrupert at yahoo.com Cc: R-help at r-project.org

Date: Tuesday,

March 17, 2009, 2:51 PM Hi


Jason Rupert wrote:

I guess no reply means there is not an

existing

package to produce

the plot?

I will post the results of my script to

hopefully help

others who are

trying to formulate the same plot.

Thanks again.

I don't know of an existing function that does

that particular 

arrangement of plots, but here's a start ...


breaks <- seq(-3.5, 3.5, by=.5) yrange <-

range(breaks)

histdata <- hist(y, breaks=breaks, plot=FALSE)

densitydata <-

density(y)

par(oma=rep(3, 4), mar=rep(0, 4))

layout(matrix(1:3, ncol=3), widths=c(.2, .1, 1),

respect=TRUE)

barplot(histdata$density, horiz=TRUE, space=0,

xlim=c(0,

max(histdata$density)*1.1), axes=FALSE)

par(new=TRUE) 

plot(densitydata$y, densitydata$x, xlim=c(0,
max(histdata$density)*1.1), ylim=yrange,

type="l", ann=FALSE,

axes=FALSE) axis(2) box() boxplot(y, ylim=yrange,

axes=FALSE) box()

 qqnorm(y, axes=FALSE, ylim=yrange, xlim=yrange,

main="") abline(0,

1) axis(1) box()


... the important bits are the call to layout() to

get the plots at

 different sizes and the numerous efforts made to

make sure that

the coordinate systems of the various plots are

coherent.

Paul

--- On Mon, 3/16/09, Jason Rupert

<jasonkrupert at yahoo.com> wrote:

From: Jason Rupert

<jasonkrupert at yahoo.com>

Subject: [R] R package

to automatically produce combination plot?

To:

R-help at r-project.org

Date: Monday, March 16, 2009, 8:14 PM By

any

chance is there an R

package that automatically produces the

plot shown

at the following

link:

http://n2.nabble.com/Can-R-produce-this-plot--td2489288.html

That is an R package to produce on plot

that has

the following: (a)

a vertically oriented histogram, (b)

associated

barplot, and (c)

quantile-quantile plot (Q-Q Plot).

This is based on a class lecture from

University

of Pennsylvania: 

stat.wharton.upenn.edu/~mcjon/stat-431/lecture-02.pdf

I am pretty confident I can put one

together, but

just wanted to

check that there does not already exist an

R

package to output such

a plot.

Thanks again.