PLOTTING BASICS

topics:

*basic plots
*maps
*images
*printing graphs







To open the graphics window In R:

>X11()


To close the graphics window:

> dev.off()

The following commands create a simple scatterplot; the semicolon is used to 
put more than one command on the line.

> X11(); plot(1:10,1:10)

View plot

Arguments which add titles or modify the appearance of the plot can be 
included in the function:


> plot(1:10,1:10,type="l",main="Straight line")

View plot

> hist(rnorm(100), main="Histogram of 100 random normal data points")

View histogram

>dev.off() 

Graphical Methods 

Plot

                    plot(x,y,...)

> xl<-"April 1975 - Dec 1975"

> X11(); par(mfrow=c(3,1))
> read.table("climate")->climate
> plot(climate[ ,1], xlab=xl, main="plot", type='l')

> plot(lowess(climate[ ,1],f=1/3),xlab=xl, main="plot using lowess", 
type='l')
# f is the   fraction of the data used for smoothing at each x point. 
# The larger the f value, the smoother  the fit.
 

> dev.off()

View plot

When only one variable is specified in the arguments to plot(), the values of 
the variable are plotted against their indices, or
against time in the case of time series data.


> X11(); plot(climate[ ,1], climate[ ,2])

> abline(lsfit(climate[,1],climate[,2]))

View plot

Barplot

                   barplot(x,names=NULL,...)

Creates a bargraph. Several options are available including verticle or 
horizontal bars and shading patterns.

names=NULL  character vector of names for the bars

> grades<-c(10,14,20,10)

> grade.names<-c('Poor','Fair','Good','Excellent')

> barplot(grades, names=grade.names, main="Barplot of exam grades")

View barplot

Pie

                   pie(x, names=NULL, explode=F, ...)

Creates a pie chart from a vector of data.

names=NULL  vector of slice labels

> pie(grades, main="Pie chart of exam grades")


View pie

Dotchart

                   dotchart(x, labels= , groups=NULL, pch="o",...)

Creates a dot chart from a vector of data. A grouping variable, and a group 
summary may be used along with other options.

labels=      vector of labels for the data values
groups=NULL  categorical variable used for splitting data into groups


 grades<- c(10,14,20,10,15,16,15,8,5,10,25,14)

 grade.names<-c('Poor','Fair','Good','Excellent')

 grade.group<-factor(c(1,1,1,1,2,2,2,2,3,3,3,3), labels=c("Test 1","Test 2", 
"Final"))

 dotchart(grades, labels = grade.names, group = grade.group)

 title(main = "Distribution of marks by test")

View dotchart

Hist

                   hist(x, nclass= , breaks= , probability=F,...)

Creates a histogram. The same options available in barplot() are available in 
hist().

nclass=        specifies the number of classes (ie.:  bars)
breaks=        vector of the break points for the bars of the histogram
probability=F  if TRUE, the histogram will be scaled as a probability density

> hist(rt(50,5), main="50 observations from a t distribution with 5 df")
# random simulations from t distribution

View histogram

Density

                   density(x, n=50, na.rm=F,...)

Returns x and y coordinates of an estimate of the probability density of the 
data.

n=50       number of equally spaced points at which to estimate the density
na.rm=F    logical flag: by default missing values generate an error message
width=     width of the window used in the computation
           the larger the value, the smoother the density

> normal<-rnorm(100)

> ndens<-density(normal, width=0.9)

> hist(normal, probability=T)

> lines(ndens)

View density plot

Specifying probability=T scales the histogram as a probability density so 
that both the line and the histogram are in the same
scale. The demo() function shows an example where the histogram was not 
scaled but rather the ylim= argument was
specified to achieve a similar effect.

Boxplot

                   boxplot(...)

Produces side by side boxplots from a number of vectors. The boxplots can be 
made to display the variability of the median, and
can have variable widths to represent differences in sample size.

 
 boxplot(split(grades, grade.group))

 
View boxplot




Pairs

              pairs(matrix, ...)

Produces all pair-wise scatter plots.

> pairs(climate)


# Here is how to create a graph in postscript 
# and save it in your Unix account (command printgraph)


X11()
x<- seq( 0,5,,300)
y1<- exp( -x)

plot( x,y1)   # plot x versus y1

dev.off() # turn off postscript device



postscript(file="ex3.ps")
x<- seq( 0,5,,300)
y1<- exp( -x)

plot( x,y1)   # plot x versus y1

dev.off() # turn off postscript device

#the postscript file is saved in your Unix directory
# use
#stat%   ghostview ex3.ps to preview it. 

# Now make some changes 
# add in   lines
y3<- sin( 2*pi*x)
plot( x,y3) 
lines(x, y3, lty=3)



# here is a interesting plot
y3<- sin( 2*pi*x)*  exp( -x) 

y4<- cos( 2*pi*x)*y1

plot( y3,y4, type="l")  # 
 ind<- y1 > max(y1)*.8    # a logical variable indicating when the 
                        #   exponential envelope is 80% of its maximum 
                         #value

points( y3[ind], y4[ind]) # now indicate these
 
polygon( c(0,y3[ind]), c(0,y4[ind]) ) # mark the area where
# the  exponential envelope is 80% of its maximum 



# Now some practice adding text

------------------------------------------------------

MAPS


>library(fields)
>US()

points( climate$lon, climate$lat, pch="x") # use period for plot character
text( climate$lon, climate$lat,  climate$elev) # label by elevation



## xlim and y lim to indicate the limits of your usa graph


US( xlim=c(-82, -73), ylim=c(32.5,41))
points( climate$lon, climate$lat, pch="o") # use period for plot character




# label by city names but make the text 1/2 the default size. 
# left justify the label

locator(1)-> hold # now click on a point in the plot with the left button
 hold # gives you the coordinates

points( hold, pch="X") # 
text( hold$x, hold$y, "Here is my favorite place", adj=1)
                # adj=1 right justify the label


----------------------------------------------------------------


GRAPHS

# Here are some methods for putting several plots on a page

x<- seq( -3,3,,80)
#creates a vector with values from -3 to 3 of length 80
y1<- dnorm( x) #  N(0,1) density 
y2<- pnorm( x) # N(0,1)  CDF

plot( x,y1, type="l")
plot( x,y2, type="l") # The problem is that the second plot erases the first

par(mfrow=c( 2,1)) # divide page into a 2X1 panel
plot( x,y1, type="l")
plot( x,y2, type="l") 
title("Standard Normal CDF")  # title wil go to the last plot drawn


More examples:

# Here is a way to view some histogram of random N(0,1) samples
par(mfrow=c( 4,4)) 
for ( k in 1:16) { plot( rnorm(50))}  # for loop  will go 16 times
#rnorm generates at random sample of N(0,1)'s


par(mfrow=c(1,2)) # this is a pleasing aspect ratio for four plots
set.seed(123)  #fix seed so that we all see the same thing
X<- matrix( rnorm(200*3), ncol=3)   # X is a matrix of random N(0,1)'s

boxplot( X[,1],X[,2], X[,3]) # SPLUS style boxplot
title("boxplot function")

hist( X[,1])
title( "hist function")



# Options in par


par( cex=.8) # make the character expansion size smaller


pairs(X) # redraw


------------------------------------------------------------

IMAGES:


# Using surface plotting 


x1<- 5:10
x2<- 1:6
outer( x1,x2,"+")-> f
f
par(mfrow=c(2,2))
image( x1,x2,f)
persp( x1,x2,f)
contour( x1,x2,f)
# an important list for surface plots
look<-  list( x=x1, y=x2,z=f) 
image(look)

----------------

SPATIAL INTERPOLATION



library(geoR)

 data(s100)
#Loads specified data sets, or list the available data sets.


     loci <- expand.grid(seq(0,1,l=31), seq(0,1,l=31))
    
#create a 31x31 grid
#interpolation at the "loci" locations":

     kc <- ksline(coords=s100$coords,data=s100$data, loc=loci,cov.pars=c(1,.5))
#spatial interpolation of our dataset s100
# at the locations in the grid "loci"


kc$predict
#predicted (interpolated) values


kc$krige.var
#variance of our interpolated values

#image graph of the interpolated values and their
#standard error at the "loci" locations:

     par(mfrow=c(1,2))
     image.kriging(kc, loc=loci, main="interpolated values")
     image.kriging(kc, loc=loci, val=sqrt(kc$krige.var),
                   main="std. errors")




Printing
---------

postscript("file.ps")
plot((1:10), col=topo.colors(64))
dev.off()



--------------------
COLOR:

The color of the created postscript file 
will be the same as the one showed in the R window 

     `col' A specification for the default plotting color. 
     `col.axis' The color to be used for axis annotation.
     `col.lab' The color to be used for x and y labels.
     `col.main' The color to be used for plot main titles.
     `col.sub' The color to be used for plot sub-titles.

Use colors() for additional information


------------------------

CONTROLING TICK INTERVALS ON AXES


par(lab=c(...))

lab=
      controls number of tick intervals and length of labels

      x= number of tick intervals on x axis (default = 5)
      y= number of tick intervals on y axis (default = 5)
      llen= length of labels on both axes (default = 7)
      eg.: lab=c(x=6, y=4, llen=9)


>par(lab=c(x=10,y=10,llen=10))
> plot(1:10,1:10)


-------------------------------


Assignment: create a vector x with equally spaced values from -4 to 4 of length 100,  divide the page into a 2X1 Panel, plot a boxplot of x and then a histogram of 50 simulated values from  a N(0,1), add a density graph to the histogram.