Transcript Slide 1

AN INTRODUCTION TO
GRAPHICS IN R
Today

Overview
– Gallery of R Graph
examples



High-Level Plotting
Functions
Low-Level Plotting
Functions
Useful functions in
conjunction with graphics
– Expand your functional
toolbox


par
Devices
(WE WILL RETURN TO THESE CONCEPTS
IN PRACTICAL EXAMPLES)

PRACTICAL/EXPERIMENTATION
Overview






One of the best things about R is the
ability to create publication-quality
graphs
Graph function syntax follows general
functional rules of the R language: the
fuzzy boundaries between dataset
management, statistical analysis, and
graphics ultimately is an advantage to
the language, but can be a little
overwhelming at first.
Corollary: Understanding graphics will
facilitate understanding R programming
in general---and it gives pretty instant
feedback
There are numerous functions available
for producing diverse output, with many
commonalities in syntax between them
There also are many ways to
accomplish a given task.
How a plotting function deals with data
often depends on the data type (e.g.
matrix, factor, vector) given to it
Ooooh….Aaaah…
A graph comprised of multiple plots
par(“mfrow”=c(2,1)) #change default behavior of plot function to plot 2 graphs per device
And more:
“multi-bivariate” visualization…
Plotting Commands




High-level: create new graphs that inherit properties of one type of plot or
another and (e.g. scatterplot) and define graphspace (e.g. x and y
boundaries).
Low-Level: additions to existing plots (e.g. lines, additional series, text,
shapes)
JUST LIKE OTHER COMMANDS IN R, PLOTTING USES FUNCTIONS THAT
TAKE 1 OR MORE ARGUMENTS, WITH PRESET ARGUMENTS THAT CAN BE
CHANGED AS DESIRED
par “commands”: default plotting parameters that can be modified to change
global behavior of subsequent commands. In essence, you are changing
your global preferences for graphing. Some of the par variables can be
temporarily changed by adding arguments to high or low level plotting
commands; in this case, the par variable is changed for that command only.
Examples of high level
plotting commands:









plot()
hist()
boxplot()
barplot()
dotchart()
pie()
qqplot()
qqnorm()
pairs()

“3D” functions
–
–
–
–
–
–
heatmap()
image()
persp()
contour()
filled.contour()
heatmap.2()
Low-level plotting
commands

points()



lines()
abline()
arrows()
segments()

rug()




…

text()
mtext()
legend()

polygon()
rectangle()

…

GRAPH COMMANDS ARE FUNCTIONS
and can call other functions







sort()
sum()
density()
lowess()
spline()
ifelse()
…
Par


A list of graphical parameters that define the
default behavior of all plot functions.
Just like other R objects, par elements are similarly
modifiable, with slightly different syntax
– e.g. par(“bg”=“lightcyan”)
– This would change the background color of all subsequent
plots to light cyan


When par elements are modified directly (as above,
this changes all subsequent plotting behavior
Some par elements can be modified from within
high and low level plotting functions. In this case,
Par parameter examples often modifiable from
within plotting functions

bg – plot background color
lty – line type (e.g. dot, dash, solid)
lwd – line width
col – color
cex – text size inside plot
mex – text size in margins

mfcol/mfrow – multiple plot option





– 2 element vector (#rows,#cols)
… many, many more (you tend to learn them as you need
them)
Add On Packages

A huge number of additional high
and low-level plotting functions
are available within add-on
packages. Genetics and social
sciences are particularly wellrepresented.
plot using the enhanced layout
capabilities of the lattice package
Statistical Function Output





Many statistical functions (regression,
cluster analysis) create special objects.
Some are special. These arguments will
automatically format graphical output in a
specific way.
e.g.
lm()
hclust()
agnes()
Devices



Specify Destination of Graphics Output
Could be windows in R
Could be files
– Not Scalable



JPG
BMP
PNG
– Scalable:



Postscript
Pdf
pictex
– Others

Win.metafile
Practical
1)
Demonstrate Tools (me)
2)
Put them to use to solve silly
problems (you)
WHO LEFT OUT THE MILK CLUB MILK?
FALSE COLOR IMAGERY
CHARACTERISTIC
SPLATTER
PATTERN
“MUG SHOT”
CRIME SCENE
(AERIAL VIEW)
OBLIQUE CARTON
PLACEMENT: EVIDENCE
OF ANTISOCIAL BEHAVIOR
PERP’s PEN?
MILK SPILL
Practical



Make sure that the .Rdata file and the .R file
are in the same folder.
In Windows, double-clicking on an .RData
file will 1) fire up R, 2) load the objects in
the file, and 3) change working directory to
the file location. Then open today’s R script.
Alternatively, one can open the R script first,
setwd(), and then use the
load("RgraphExamples.RData") command
in the script to load the objects for analysis