R Graphics

• Chart area

• Chart area
• Plot area

• Chart area
• Plot area
• Data representation

• Chart area
• Plot area
• Data representation
• Axis (scaling, labeling)

• Chart area
• Plot area
• Data representation
• Axis (scaling, labeling)
• Figure title

• Chart area
• Plot area
• Data representation
• Axis (scaling, labeling)
• Figure title
• Legend

• Do not use pie chart
• Do not use 3D (never)
• Use consistent colors

• Do not use pie chart
• Do not use 3D (never)
• Use consistent colors
• Do prefer this representation

• Do not use more than 6 colors
• Do not use high contrast color

• Do not use more than 6 colors
• Do not use high contrast color
• Sometimes sizes and symbols are better

• Write textual informations horizontally
• Do not use distracting elements
• Do not add chart junk

• Write textual informations horizontally
• Do not use distracting elements
• Do not add chart junk
• Think about the Data-Ink ratio (Tufte, 1983)

• Base package
• S-like plotting functions
• Contains the famous function plot()
• And a lot of well-known functions:
  • boxplot(), barplot(), hist()
  • lines(), points(), legend()
  • etc.

• An alternative set of graphical functions
• Well-suited for developers
• ggplot2 is based on this package

• Based on the grid package
• High-level system inspired by Trellis graphics
• Specialized on multivariate data
• and multipanel figures

• Also based on the grid package
• A complete plotting system for R
• Based on the Grammar of Graphics
• But introduces its own syntax

• and requires a long time to master it

• See the QCBS workshop on ggplot2

• Based on the graphics package
• lots of specialized plots (i.e. polar plots)
• axis, labeling and color scaling functions

• Based on the graphics package
• Enhanced versions of standard plots ( boxplot2)
• some extra functions (e.g. Venn diagram)

• par() returns their values
• par() allows changing the values

• Random data with no particular sense
• Three variables:
  • x and y: quantitative variables
  • z: qualitative variable (factor)

• The default plot
• Quite ugly, isn't it?

• Now we remove all component
• We make an empty graph

• First let's remove the box
• with the argument bty (default: 'o')

• Now let's remove the textual annotation
• with the argument ann (default: 'TRUE')

• Let's remove the x-axis
• with the argument xaxt (default: 's')

• And the y-axis
• with the argument yaxt (default: 's')

• Using axes=FALSE is the same as:
• bty='n'+xaxt='n'+yaxt='n'

• Finally let's remove data
• with the argument type (default: 'p')

• Finally let's remove data
• with the argument type (default: 'p')

• In an empty plot, visual information is not displayed but the graph is defined in the window
• It is now possible to use low-level plotting functions such as points() or axis()

• Let's use the function points()
• It shares some arguments with plot()

• We can customize the points with:
  • cex, the size
  • col, the color
  • pch, the symbol

• Some symbols have two colors:
  • col: the border color,
  • bg: the background color
• This is the case for pch = 21 to 25

• Four functions allow to plot lines:
  • points()
  • lines()
  • abline()
  • segments()
• Let's examplify with a linear regression

• First, let's try the function abline()
• with the first way

• The second way is to specify model parameters

• The function abline() allows to draw horizontal and vertical lines

• Now take a look at the functions lines() and points()
• But first, we are going to predict the model on new data

• Let's add model regression with the functions lines() and points()

• We can customize the lines with:
  • lwd, the line width
  • col, the line color
  • lty, the line type

• To add a polygon, the function is polygon()
• A special, the rectangle can be drawn with rect()
• Let's predict again the model, but this time with the standard error

• We are going to add the error envelope with the function polygon()
• So, let's calculate the coordinates of this envelope

• Let's add model error envelope

• We can customize the polygon with:
  • border, the border color
  • col, the background color
  • lwd, the border width

• Finally

• rect() is appropriated when you want to add draw rectangle
• Here is an example

• You also can customize the rectangle

• Let's see now how to add text
• First let's add a main title with the function title()

• What about adding text in the plot area?
• We will use the function text()
• Here is a first example:

• another example

• Let's customize a little the text with:
  • cex, the size
  • col, the color
  • font, the font (bold, italic, etc.)
  • family, the typeface

• Finally, let's customize the orientation and position
  • srt, the rotation angle
  • pos, the position from coordinates

• But the function text() can't add text outside the plot area
• except if par(xpd=TRUE)
• We will use the function mtext()

• You can define colors in the RGB system with rgb()
• You specify value between 0 to 1 for each primary colors
• For example:

red    <- rgb(red = 1,   green = 0,   blue = 0)
yellow <- rgb(red = 1,   green = 1,   blue = 0)
gray1  <- rgb(red = 0.5, green = 0.5, blue = 0.5)

• You can define colors in the RGB system with rgb()
• You specify value between 0 to 1 for each primary colors
• For example:

red    <- rgb(red = 1,   green = 0,   blue = 0)
yellow <- rgb(red = 1,   green = 1,   blue = 0)
gray1  <- rgb(red = 0.5, green = 0.5, blue = 0.5)

• Let's see an application

## Empty plot
plot(x = dat$x, y = dat$y, ann = FALSE,
     bty = 'n', type = 'n')

## Transparent red
redp <- rgb(red = 1, green = 0, blue = 0,
            alpha = .5)

## Adding points
points(x = dat$x, y = dat$y, col = redp,
       pch = 19, cex = 4)

• You can also define colors in the hexadecimal system
• Each primary colors is define by two values varying from 0 to 9 and A to F
• For example:

red    <- '#FF0000'
yellow <- '#FFFF00'
gray1  <- '#888888'

• You can also define colors in the hexadecimal system
• Each primary colors is define by two values varying from 0 to 9 and A to F
• For example:

red    <- '#FF0000'
yellow <- '#FFFF00'
gray1  <- '#888888'

• The function axis() allows to add axis
• Here is an example of usage

## Empty plot
plot(x = dat$x, y = dat$y, pch = 19)

## Adding top-axis
axis(side = 3, at = seq(0.6, 1.4, by = 0.1),
     labels = seq(0.6, 1.4, by = 0.1), las = 1)

## Adding right-axis
axis(side = 4, at = seq(0.8, 1.2, by = 0.1),
     labels = format(seq(0.8, 1.2, by = 0.1)),
     las = 2)

• To change the figure margins you have to change the values of the parameter mar in the par()
• The order is the follow: bottom, left, top and right
• For example:

par(mar = c(4, 4, 4, 4))

• Use the dataset in 'iris.txt' (available on    in the 'data' folder)
• Using the dataset iris.txt (Dropbox)

• mfrow and mfcol in par()

par(mfrow=c(2,2))

or

par(mfcol=c(2,2))

• mfrow and mfcol in par()
• split.screen()

split.screen(c(1, 2))
split.screen(c(3, 1), screen = 2)

• mfrow and mfcol in par()
• split.screen()
• layout()

mat_lay <- matrix(c(1,2,4,1,3,4),nrow=3)
layout(mat_lay)

mat_lay <- matrix(c(1,2,4,1,3,4), nrow=3)
layout(mat_lay)

##      [,1] [,2]
## [1,]    1    1
## [2,]    2    3
## [3,]    4    4

mat_lay <- matrix(c(0,2,2,1,3,3,1,4,0), nrow=3)
layout(mat_lay)

##      [,1] [,2] [,3]
## [1,]    0    1    1
## [2,]    2    3    4
## [3,]    2    3    0

mat_lay <- matrix(c(0,2,2,1,3,3,1,4,0),nrow=3)
layout(mat_lay, widths=c(.25,1,1))

mat_lay <- matrix(c(0,2,2,1,3,3,1,4,0),nrow=3)
layout(mat_lay, widths=c(.25,1,1),
  heights=c(.25,1,.25))

• You must call new=TRUE
• AND specifying fig in par()

1. create your first plot;

  plot(...)

• You must call new=TRUE
• AND specifying fig in par()

1. create your first plot;
2. use par();

  plot(...)
  par(new=TRUE, fig=c(0.5,1,0.5,1))

• You must call new=TRUE
• AND specifying fig in par()

1. create your first plot;
2. use par();

3. add your embedded plot;

   plot(...)
   par(new=TRUE, fig=c(0.5,1,0.5,1))
   plot(...)
   

• bmp(), jpeg(), png(), tiff()

?jpeg

• use them:

png(filename, width=480, height=480)
...
dev.off()

png(filename, width=1440, height=1440)
...
dev.off()

jpeg(filename, res=72,
  pointsize=12, width=480, height=480)
...
dev.off()

• pointsize=12
• res=72
• cex=1

png(filename, res=144)
...
dev.off()

png(filename, res=144,
  height=7, width=7, unit="in")
...
dev.off()

png(filename, res=288,
   height=7, width=7, unit="in")
...
dev.off()

png(filename, res=288,
  height=7, width=7, unit="in")
...
dev.off()

png(filename, res=288,
  height=2*7, width=7, unit="in")
...
dev.off()

pdf(fil, pointsize=12,
  height=2*7, width=7)
...
dev.off()

svg(filename, pointsize=12,
  height=2*7, width=7)
...
dev.off()