R help - Oct 2008 - how to plot the histogram and the curve in the same graph

i want to plot the histogram and the curve in the same graph.if i have a set
of data ,i plot the histogram and also want to see what distribution it
was.So i want to plot the curve to know what distribution it like.
-- 
View this message in context:
http://www.nabble.com/how-to-plot-the-histogram-and-the-curve-in--the-same-graph-tp20082506p20082506.html
Sent from the R help mailing list archive at Nabble.com.

leo_wa <kwngai6022 <at> hotmail.com> writes:
> 
> i want to plot the histogram and the curve in the same graph.if i have a 
> set  of data ,i plot the histogram and also want to see what distribution 
> it  was.So i want to plot the curve to know what distribution it like.
See the example under histogram/lattice. Also have a look at Deepayan's 
book-site.

http://lmdvr.r-forge.r-project.org/figures/figures.html

Dieter

leo_wa <kwngai6022 <at> hotmail.com> writes:
> 
> 
> i want to plot the histogram and the curve in the same graph.if i have a
set
> of data ,i plot the histogram and also want to see what distribution it
> was.So i want to plot the curve to know what distribution it like.

You will find below an example using only basic plotting functions. 
The created function (called histplot) plots a histogram of the data, 
along with a density kernel estimate of the distribution and 
(if asked by option ncurve=T) the normal distribution 
(with mean and sd computed from the original data). 

Hope this will help

Matthieu


#generating random values to be plotted
dat <- rnorm(100)

#plotting function 
histplot <- function(dat, breaks="Sturges", ncurve=TRUE, ...)
{
	#compute the histogram and density of "dat"
	hdat <- hist(dat, breaks=breaks, plot=F)
	ddat <- density(dat)
	
	#compute the xlim and ylim of the plot
	# i.e. the min and max of the different superimposed 
	#plots (hist, density and normal curves)
	xlim <- range(ddat$x)
	if(ncurve)
	{
		#max of the normal curve		
		maxnorm <- pnorm(mean(dat), mean=mean(dat), sd=sd(dat))
		ylim <- c(0,  max(hdat$density,ddat$y,maxnorm))
	}
	else
	{
		ylim <- c(0,  max(hdat$density,ddat$y))
	}
	
	#plotting 
	plot(hdat,
		freq=F,
		xlim=xlim, ylim=ylim, ...)
	lines(ddat)
	if (ncurve) curve(dnorm(x, mean=mean(dat), sd=(sd(dat))), 
				lty=3, add=TRUE)
}

#usage
histplot(dat)
histplot(dat, ncurve=F)
histplot(dat, col="blue") #arguments are passed to the hist plotting
function

leo_wa wrote:
> i want to plot the histogram and the curve in the same graph.if i have a
set
> of data ,i plot the histogram and also want to see what distribution it
> was.So i want to plot the curve to know what distribution it like.
>   
>   
To draw the curve and the distribution you should have an idea about the 
distribution.
You cann't just draw the histogram and expect R to make a curve of the 
best distribution to fit that histogram.
But you can plot a curve of a kernel density.

x <- rnorm(1000,5,3)
library(MASS)
(x.normfit <- fitdistr(x,"normal"))
hist(x,prob=TRUE)
lines(density(x,na.rm=TRUE),col="red") # kernel density
curve(dnorm(x,mean= x.normfit$estimate[1],sd= 
x.normfit$estimate[2]),col="blue",add=TRUE) #maximum likelihood
estimate

Rub?n

It is not clear what curve you want to plot, but this code may help get you
started (the final plot will probably need the x and y limits expanded and maybe
other options changed):

x <- rnorm(100, rep( c(100,105), c(75,25)), 2 )

hist(x, probability=TRUE)
lines( density(x), col='green' )
curve( dnorm(x, mean=mean(x), sd=sd(x)), col='blue', add=TRUE )

see the help for hist, lines, and curve for details.

Hope this helps,

--
Gregory (Greg) L. Snow Ph.D.
Statistical Data Center
Intermountain Healthcare
greg.snow at imail.org
801.408.8111

> -----Original Message-----
> From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-
> project.org] On Behalf Of leo_wa
> Sent: Monday, October 20, 2008 8:47 PM
> To: r-help at r-project.org
> Subject: [R] how to plot the histogram and the curve in the same graph
>
>
> i want to plot the histogram and the curve in the same graph.if i have
> a set
> of data ,i plot the histogram and also want to see what distribution it
> was.So i want to plot the curve to know what distribution it like.
> --
> View this message in context: http://www.nabble.com/how-to-plot-the-
> histogram-and-the-curve-in--the-same-graph-tp20082506p20082506.html
> Sent from the R help mailing list archive at Nabble.com.
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-
> guide.html
> and provide commented, minimal, self-contained, reproducible code.

Here is some code that more or less does what you want -- not a true 
histogram, but all else is in place.
Feel free to modify, give away, etc. No copyright ;-)


myhist<-function(data,breaks="Sturges",xlabel="data 
value",ylabel='counts',curve=TRUE, title="myhist.R made this
fake
histogram",sdf=5)
{
sudh<-hist(data,plot=FALSE,breaks=breaks)

#
# Note: the literature says a histogram relates count density to box 
volume, so
# technically this function makes a bar chart, not a histogram.  Whatever.
#
# get current graph dims and calculate  a nice bar width
pdim<-par('din')
barwd<-pdim[1]*2
# find midpoints of histogram cells IDIOT: hist() does this for you
#mids<-breaks[1:(length(breaks)-1)] +diff(breaks)/2
# get max of spline-- will almost always be > data
#smoo<-spline(sudh$breaks[1:length(sudh$breaks)-1],sudh$counts)
smoo<-spline(sudh$mids,sudh$counts)
#this was dumb: only plot smoo if curve=TRUE,
# labels and stuff  are hairy if merge a PLOT inside and outside IF
#subtle: to get same graph size, need to set ymax for plot to max of 
spline(y)
# I think usr=c(min(smoo$x),max(sudh$breaks),0,max(smoo$y)) works
if(curve) {
	plot(smoo$x, 
smoo$y,xlab=xlabel,ylab=ylabel,'l',col='green',usr=c(min(smoo$x),max(sudh$breaks),0,max(smoo$y)))
	#now plot the bars
	
	lines (sudh$mids,sudh$counts,type='h',lwd=barwd,lend=1)
	}
else {
	#just plot the hist, not the spline
#plot(sudh$breaks[1:length(sudh$breaks)-1],sudh$counts,type='h',xlab=xlabel,ylab=ylabel,lwd=barwd,lend=1)
plot(sudh$mids,sudh$counts,type='h',xlab=xlabel,ylab=ylabel,lwd=barwd,lend=1)
#lines(sudh$breaks[1:length(sudh$breaks)-1]/60,sudh$counts,col='red')
	}
title(main=title)
histstuff<-list(input=data, histogram=sudh, spline=smoo)
return(invisible(histstuff))
}