knitr::kable(courses)

Aesthetics can be defined:

• At plot level (ggplot())
  • by mapping (aes()) to data
    • inherited by all layers
• At layer level (geom_..())
  • by mapping (aes()) to data
  • at a fixed value (not scaled)
  • apply to the layer only
  • can overwrite those defined at plot level

ggplot(courses,aes(y=course,x=credits))+
  geom_bar(stat="identity")+
  geom_text(aes(label=paste(credits,"cfu")),hjust=0)

No legend is shown for the fixed value.

ggplot(courses,aes(y=course,x=credits))+
  geom_bar(stat="identity")+
  geom_text(aes(label=paste(credits,"cfu")),
            color="white", hjust=1)

Aesthetics mapped to data (aes()) are scaled and produce a legend (guide)

ggplot(courses,aes(y=course,x=credits))+
  geom_bar(aes(fill=sem),stat="identity")+
  geom_text(aes(label=paste(credits,"cfu")),
            hjust=1, color="white")

1. series are stored as distinct variables,
   • each variable is mapped in a different layer
2. series are stored as the same value variable, and another type variable tells them apart – usualy factor –
   • a single layer is used and series are separated by
     • mapping type variable to a visual or group, or/and
     • faceting (small multiples) by type

Case 1 may be trasformed into case 2 to via pivot_longer() and viceversa through pivot_wider()

• layers are overlapped and independent, so upper (later layers) cover lower ones
• no legend is produced (because fill aesthetics is not scaled)

ggplot(courses,aes(y=course))+
  geom_bar(aes(x=study_hours),stat="identity",fill="seagreen")+
  geom_bar(aes(x=lecture_hours),stat="identity",fill="orange")

courses_long <- courses %>% pivot_longer(
  ends_with("_hours"), names_to="type", values_to="hours") 
knitr::kable(courses_long)

• unique layers, bars are automatically stacked not to overlap
• a legend is produced by the (implicit) scale_fill_discrete

ggplot(courses_long, aes(y=course,x=hours,fill=type))+
  geom_bar(stat="identity")

• unique layers, bars are explicitly dodged
• color scaled by scale_fill_manual

ggplot(courses_long, aes(y=course,x=hours,fill=type))+
  geom_bar(stat="identity",position="dodge") +
  scale_fill_manual(values=c("orange","seagreen"))

Each facet contains a separate series

ggplot(courses_long, aes(y=course,x=hours,fill=type))+
  geom_bar(stat="identity")+
  scale_fill_manual(values=c("orange","seagreen")) +
  facet_wrap(type~.)

Data can be defined:

• At plot level (ggplot())
  • available to all layers
• At layer level (geom_..())
  • parameter data
  • available to that layer only

average <- courses_long %>% group_by(type) %>% 
           summarize(hours=mean(hours))
ggplot(courses_long, aes(y=course,x=hours))+
  geom_bar(stat="identity") +
  geom_vline(aes(xintercept=hours),data=average,color="red") +
  facet_wrap(type~.)

rightmost <- courses_long %>% group_by(type) %>% 
          summarize(course=last(course),hours=last(hours))
ggplot(courses_long, aes(x=course,y=hours,color=type))+
  geom_line(aes(group=type))+
  scale_color_manual(values=c("orange","seagreen"),guide=F) +
  geom_text(aes(label=type),data=rightmost, 
            hjust=0.5,vjust=1.5,show.legend=FALSE)

Text annotation layer is created with geom_text()and geom_label() with the following aesthetics:

• label: the text label
• hjust : horizontal alignment
  • "left", "middle", "right"
• vjust: vertical alignment
  • "top", "center", "bottom"
• family: font family
• fontface: type of font, e.g. “bold”, “italic”

Note: e.g., "left" means align to the left of the reference position

• geom_path() obeys the order of points in the data
• geom_line() sorts the points by x

Use geom_bar(stat="identity"), with aethetics:

• position: "stack", "dodge", "fill"

geom_smooth() provides a least square interpolation of points.

set.seed(1793); data.frame(x=1:20,y=rnorm(20)) %>%
  ggplot(aes(x,y))+geom_point()+geom_smooth()

Position scales are scale_x/y..

• _continuous() linear, with variations:
  • _log10()
  • _sqrt()
  • _reverse()
• _discrete()
• _date() with variations
  • _datetime()
  • _time()

Parameters:

• name the name/label of the axis
• breaks the breaks (ticks) of the axis, a vector or a function
  • also minor_breaks for continuous and date scales
  • also date_breaks set distance among breaks, e.g. 2 days
• labels the labels, a vector or a function
  • can use label_ functions in scales package e.g. label_percent()

• limits: set the limits (or list of values for discrete scale)
• expand: define how much space is added at extremes of axis
  • use function expansion()
    • mult: multiplicative expansion, default: 0.05 for continuous
    • add: additive expansion, default: 0.6 for discrete
    • both can be 1 or 2 elements long
    expand=expansion(mult=c(0.05,0.2))
    adds 5% to the left and 20% to the right

Works for both color and fill aesthetics.

• _manual, assign levels to color names in values
• _gradient, define a color gradient from low to high
• _brewer picks from a colorbewer predefined palette
  • type: "seq", "div", "qual"
  • palette: names (e.g., “Greens”) or number
• _viridis_b|c|d binned,continuous, discrete predefined palettes
  • option: “A” to “E”

• linetype type of line for geom_line() and geom_path()

• Maps to at most 6 different levels of a discrete variable

• Symbols can be selected with scale_shape_manual()

  • 0 to 14 are outlined shapes
  • 15 to 20 are solid shapes
  • 21 to 25 are filled shapes

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