R Programming/Graphics
From Wikibooks, the open-content textbooks collection
Contents |
[edit] Basics
[edit] plot
- The main function is plot.
- Plot opens a graph window and draw a graphics. A very basic example would be the following :
plot(0)
- Plot deals with datasets, one continuous variable, two continuous variable, one discrete variable, etc.
- One continuous variable : creates a time series plot
# Example with the anscombe dataset (included in the datasets package) plot(anscombe$x1)
-
- Two continous variables : creates a scatterplot. The first term is one the x-axis and the second one on the y-axis.
plot(anscombe$x1,anscombe$x2)
-
- one dataset : create a scatterplot matrix
plot(anscombe)
[edit] type
The type can be
- n for none
- p for points
- l for lines
- h for histogram-like
- s/S for steps
- see ?plot for more.
[edit] Symbols
- Symbols can be changed using the pch option which takes integer values between 0 and 25.
plot(x1,x2,pch=0)
- Note that symbols can also use a vector of integers.
In that case the symbol will be different for each observation.
plot(x1,x2,pch=c(1:10))
- If you want to have a broad look at all symbols, you can do a simple loop over pch :
par(mfrow = c(4,3))
for (k in 0:11){
plot(x1,x2,pch=k)
}
[edit] Axes
- The default output print the axes.
- You can remove them :
attach(anscombe) plot(x1,x2,axes=FALSE)
- You can also change the axes
plot(x1,x2,axes=F) axis(1,col="red",col.axis="blue",font.axis=3) axis(2,col="red",col.axis="blue",font.axis=2,las=2)
[edit] Size
- cex change the size of the output
[edit] Using graphics parameters
- The par() command defines most of the parameters of a plot.
If you just type par() in R, you will se the whole list of default arguments
[edit] Colors
- Colors in R
- type :
colors()
[edit] Exporting your graphs
- CTRL + W : copy the current graph as a Window Metafile in the clipboard. You can paste it in another application.
- CTRL + C : copy the current graph as a Bitmap file in the clipboard.
- The Windows Metafile is richer than the Bitmap format.
- Use the contextual menu (Right clic on Windows and Linux or Control + Clic on Mac)
- Export of the plot to pdf is easily performed by adding pdf("filename.pdf") prior to the plotting, and dev.off() after the plotting.
- Export to png can similarly be performed by calling png("filename.png") prior to the plotting, and dev.off() after the plotting.
- Use the savePlot() function to save existing graphics
Here is an example of export to png format:
png("r_plot.png", width = 420, height = 340)
…
dev.off()
Here is an example of export to ps format:
postscript(file="W:/…/graph1.ps",horizontal=F, pagecentre=F,paper="special",width=8.33,height=5.56) … dev.off()
You can export to SVG:
#R version 2.3, www.r-project.org
library("RSvgDevice")
devSVG("histogram.svg", width = 5, height = 4)
set.seed(50)
par(mar=c(3,2,1,1),yaxs="i",mgp=c(1.8,.9,0))
hist(rnorm(100),col="lightblue",main="",xlab="",ylab="",ylim=c(0,28))
box()
dev.off()
Here is the output :
[edit] Topics
[edit] Time series plot
[edit] Scatter plot
The first variable in the horizontal axis and the second one in the vertical axis
plot(x,y)
[edit] Overlaying scatterplots
If one wants to overlay two scatterplot, he can use the new=FALSE option in the par() command. Here is an example :
plot(effec_passe,p_3,type="p",ylab="Probabilité",xlab="Effectif d'origine") par(new=TRUE,col="red") # creates a new graphic plot(effec_passe,p_prev_3,axes=F,type="p",ylab="",xlab="") par(new=FALSE,col="black") # does not create a new graphic
[edit] Regression line
[edit] Bar charts
[edit] Dot chart
[edit] Standard error bar chart
- Standard error bar chart are very useful to plot estimates with information on confidence interval.
- The Hmisc package has an errbar() function.
- NOTE : The default output in errbar() changed between R version 2.8.1 and R version 2.9.2. Axis are not displayed by default anymore.
- See also the coefplot() function in Gelman and Hill's "arm" package
[edit] Analytic Graphics
Alpha <- .3
S <- .16
N <- .005
Gamma <- .015
Mu <- .03
f1 <- function(k.e) S * k.e^Alpha
f2 <- function(k.e) (N + Mu + Gamma)*k.e
plot(0,type="n",xlim=c(0,8),ylim=c(0,.5),ylab="",xlab=(expression(k^e)))
curve(f1, add = T)
curve(f2, add = T)
text(7,f1(7), expression(s * k^e^alpha ))
text(7,f2(7), expression(paste("(",n + mu + gamma,")","*",k^e)))
[edit] Miscellanous
# Setup empty plot window
plot(0, type = "n", main = "Plot Title", xlab = "X axis", ylab = "Y axis", xlim = c(1970,2000), ylim = c(0,100))
grid() # add grid
lines(data[,1], data[,2], lty = 1) # draw first dataset
lines(data[,1], data[,3], lty = 2) # draw second dataset
legend("topright", c("first dataset", "second dataset"), lty = c(1, 2) )
[edit] Maps
- Geograpic Maps in R (Revolution Computing Blog)
- Global Administrative Areas
- Andy Egger's googlemap.r function
[edit] References
- A list of all Wikimedia graphics created with R
- Tables 2 Graphs
- R Graph Gallery
- See ggplot2 package
- See Lattice package
