• Why ggplot2
• Component of Grammar of Graphic
• Basic Structure
• Construct Plot Layer by Layer
• Scales, Position, Facets and Themes
• An Example

• Grammar of graphic
• Both quick and complex plot
• Easy to "create" plot
• Nice aesthetic settings
• Great docummentation and active mailing list

Things you cannot do:

• 3-dimensional graphics (rgl)
• Interactive graphics (ggvis, plotly)

• 2017 Airbnb data of New York City

• 43234 observations and 16 variables

• It consists of features of each Airbnb room in NYC such as price, review per month, name of neighbourhood, name of borough, latitude and longitude

• Simple cleansing

ggplot(data, aes(x = ,y = )) + layers + additional elements

ggplot(Airbnb, aes(x = reviews_per_month, y = price))

• Display the data or the statistical summaries of the data

• Mainly use geom_xxx() function

• An alternative way is stat_xxx() function

• A plot must have at least one geom or stat function. there is no upper limit. You can add a layer to a plot using the + operator

• All kinds of geom function and stat function: ggplot2 cheat sheet

ggplot(Airbnb, aes(price, reviews_per_month)) +
  geom_point(size = 0.1) +
  facet_grid(~room_type)

• Describe how variables are mapped to visual properties

• aes()

• Specifying the Aesthetics in the Plot(ggplot()) or in the Layers(geom_xxx() or stat_xxx())

• Aesthetic Mapping can consists of Position (i.e., on the x and y axes), color (“outside” color), fill (“inside” color), shape (of points), linetype and size, etc.

ggplot(Airbnb, aes(x = price)) +
  geom_histogram(bins = 40, aes(color = room_type), fill = "grey") 

ggplot(Airbnb, aes(x = price)) +
  geom_histogram(bins = 40, aes(fill = room_type), color = "grey") 

ggplot(Airbnb) +
  geom_violin(aes(neighbourhood_group, price), colour = 'blue')

ggplot(Airbnb) +
  geom_violin(aes(neighbourhood_group, price), colour = 'blue') +
  geom_boxplot(aes(neighbourhood_group, price), width = 0.16, 
               outlier.size = 0, notch = TRUE)

w1 <- ggplot(Airbnb) +
  geom_bar(aes(x = neighbourhood_group, fill = room_type))
w1

w1 + scale_x_discrete(name = 'Borough') +
  scale_y_continuous(name = 'Number of rooms',
                     breaks = seq(0,20000,4000)) +
  scale_fill_grey(name = 'Room type')

• Control appearance of non-data elements

• Titles, tick marks and labels

• Legends appearance

• Overall look

• theme()

• Facetting generates small multiples each showing a different subset of the data.

• facet_grid() and facet_wrap()

• facet_grid() is fundamentally 2d, being made up of two independent components.

• facet_wrap() is 1d, but wrapped into 2d to save space

ggplot(Airbnb, aes(price)) +
  geom_histogram(bins = 40) +
  facet_wrap(~neighbourhood_group)

ggplot(Airbnb, aes(price)) +
  geom_histogram(bins = 40) +
  facet_grid(room_type ~ neighbourhood_group)

How to construct this plot?

Airbnb_sub1 <- aggregate(Airbnb[ ,c(10,14)], 
  list(Airbnb$neighbourhood, Airbnb$neighbourhood_group,
       Airbnb$room_type), mean)
neighbourhood_count <- data.frame(table(Airbnb$neighbourhood,
                  Airbnb$neighbourhood_group, Airbnb$room_type))
neighbourhood_count <- 
      neighbourhood_count[which(neighbourhood_count$Freq != 0), ]
Airbnb_sub2 <- merge(Airbnb_sub1, neighbourhood_count, 
      by.x =c( 'Group.1', 'Group.2', 'Group.3'), 
      by.y = c('Var1', 'Var2', 'Var3'), all.x = TRUE)
      names(Airbnb_sub2) <- c('neighbourhood', 'neighbourhood_group',
                'room_type', 'average_price', 'average_rpm', 'count') 
good_neighbourhood <- 
  subset(Airbnb_sub2, Airbnb_sub2$average_rpm >= 3)
good_neighbourhood <- 
  subset(good_neighbourhood, good_neighbourhood$average_price <=150)

p <- ggplot(Airbnb_sub2, aes(average_price, average_rpm)) +
  geom_point(aes(color = neighbourhood_group, size = count)) +
  facet_wrap(~room_type)
p

• Change the point shape to open circle

• Add horizontal and vertical line

• Add name of those 'good neighbourhoods'

• Title, label axes, legend

• Change theme

p1 <- ggplot(Airbnb_sub2, aes(average_price, average_rpm)) +
  geom_point(aes(color = neighbourhood_group, size = count),
             shape = 1, stroke = 1) +
  facet_wrap(~room_type)
p1

p2<- p1 +  geom_vline(aes(xintercept = 150), color = 'grey') +
  geom_hline(aes(yintercept = 3), color = 'grey')
p2

p3 <- p2 + geom_text(aes(label = neighbourhood), 
           data = good_neighbourhood, size = 2.5)
p3

library(ggrepel)
p3 <- p2 + geom_text_repel(aes(label = neighbourhood), 
           data = good_neighbourhood, size = 2.5)
p3

p4 <- p3 + theme_minimal()
p4

p5 <- p4 +
  scale_x_continuous(name = 'Average Price',
                     breaks = seq(0, 400, 100)) +
  scale_y_continuous(name = 'Average Reviews Per Month', 
                     breaks = c(1:8)) +
  scale_color_discrete(name = '') +
  scale_size_continuous(name = '') +
  labs(title = 'Neibourhood in NYC Airbnb') +
  theme(legend.position = 'top',
        legend.text = element_text(size = 8, color = 'gray10'),
    plot.title = element_text(size = 15, face = 'bold', hjust = 0.5))

p5

• Wickham, H. (2016). ggplot2: elegant graphics for data analysis. Springer.

• ggplot2 cheat sheet

• ggplot2 workshop

• Top 50 ggplot2 Visualizations