The ggplot2 is packagefor producing statistical graphics.

• ggplot2 is based on a grammar
  • allows composing graphs as combination of independent components
• ggplot2 takes care of fiddly details
  • defaults let produce publication-quality graphics in seconds
• ggplot2 is designed to work incrementally
  • start raw data, then add layers of annotations and statistical summaries

Plot is composed of:

• data the information to be visualized (data frame)
• mapping of data onto aesthetic attributes
  • layer
    • geometric elements (geom)
    • statistical transformations (stat)
  • scale: maps data to attributes (e.g., color, size ..)
  • coord system: maps data coordinates to the plane
  • facet: breaks up the plot as small multiples
  • theme: provide support elements and controls details

Any ggplot2 plot has three key components:

• the data
• aesthetic mappings
  • maps data variables to aesthetics features
  • coordinates or attributes
• visual layer (at least one)
  • define the visual object
  • maps aesthetics features to geometric properties

ggplot(series, aes(x=i,y=fibonacci))+geom_point()

ggplot(series, aes(x=i,y=fibonacci))+geom_point()

• series : defines the data to be used
• aes(x=i,y=fibonacci) : maps data to visual characteristics
  • i and fibonacci to the x and y coordinates respectively
  • cartesian coordinates are implied by default
  • linear scales implied
• geom_point() : defines a layer that map data to points
  • shape, color, size of points are implied by default

• Scale depends on the type of aesthetics
  • for position (x, y) is by default a simple linear scale
  • for other types of aesthetics may vary

Both scale and coordinates have (implicit) defaults:

• the default scale depends on
  • the specific aestethics
  • the type of the variable
• the default coordinate system is coord_cartesian()
  • another option is coor_polar()

ggplot(series, aes(x=factor(i),y=fibonacci))+geom_point()

A factor is mapped to equidistanced slots along the axis

ggplot(series, aes(x=i,y=linear))+geom_point()+
      coord_polar()

x maps to \(\theta\) (with max(x) \(\rightarrow 2\pi\)) and y maps to \(\rho\)

ggplot(series, aes(x=i,y=square))+geom_point()+
      scale_y_log10(minor_breaks=c(1:10,1:10*10))

Applied a log scale to the position y

Aesthetics include:

• position (x, y)
• grouping (group)
• other:
  • color : line or simbol color
  • fill : area fill color
  • shape : type of shape
  • size : size of the object

ggplot(series %>% mutate( mag = fibonacci %/% 10), 
       aes(x=i, y=fibonacci, color=mag))+ geom_point()

A gradient scale is used for a continuous (numeric) variable

ggplot(series%>%mutate( mag = fibonacci %/% 10), 
       aes(x=i, y=fibonacci, color=factor(mag)))+ geom_point()

Discrete color scale is used for a factor variable

For each aesthetics type a few scales are provided:

• scale_x_.., scale_y_..
• scale_color_..
• scale_fill_..
• scale_shape_..
• scale_size_..

ggplot(series%>%mutate( mag = fibonacci %/% 10), 
       aes(x=i, y=fibonacci, color=mag))+ 
       scale_color_gradient(low="blue",high="gold")+ 
       geom_point()

Geometry function add new layers

• geom_point() : draw points
• geom_line() : draw lines connecting positions
• geom_text() and geom_label() : write a text or label
• geom_area() : draw a filled area

Layers are drawn in order of declaration, with the latest on top.

The order of all other statements is irrelevant.

ggplot(series, aes(x=i, y=fibonacci))+ 
       geom_line()

ggplot(series, aes(x=i, y=fibonacci, label=fibonacci))+ 
       geom_line() + geom_label()

A few geometries perform a transformation befor mapping to an object

• geom_bar() : compute frequencies of discrete variables
• geom_histogram() : compute frequencies of bins of continuous vars
• geom_boxplot() : compute boxplot
• geom_violin(): compute a violin plot

ch = strsplit("All along the watchtower",c())[[1]]
ggplot(data.frame(ch=ch), aes(x=ch))+ geom_bar()

ggplot(series, aes(x=factor(i),y=fibonacci))+ 
      geom_bar(stat="identity")

Conventional bar plot uses stat identity (instead of count)

ggplot(series, aes(x=fibonacci))+ 
      geom_histogram(binwidth=1)

ggplot(series, aes(x=fibonacci))+ 
      geom_boxplot()

The support elements and default visual features are defined by a theme

• theme_classic() : similar to base functions
• theme_gray() : the default theme (gray background)
• theme_bw() : same as default but with white backgound
• theme_light() : same as bw but with lighter lines
• theme_dark() : dark gray background
• theme_minimal() : minimalistic theme
• theme_void() : no supporting elements

ggplot(series, aes(x=factor(i),y=fibonacci))+geom_point()+
    theme_minimal()

• Hadley Wickham, Danielle Navarro, and Thomas Lin Pedersen. “ggplot2: Elegant Graphics for Data Analysis”, in-prograss
  • https://ggplot2-book.org/
• Winston Chang, “R Graphics Cookbook” O’Reilly, 2013