Summarising Anscombe’s quartet

Tmisc::quartet |>
  group_by(set) |>
  summarise(
    mean_x = mean(x), 
    mean_y = mean(y),
    sd_x = sd(x),
    sd_y = sd(y),
    r = cor(x, y)
  )|> 
  gt::gt() # print as table

Visualizing Anscombe’s quartet

ggplot2 \(\in\) tidyverse

• ggplot2 is tidyverse’s data visualization package.

• gg in “ggplot2” stands for Grammar of Graphics.

• Inspired by the book Grammar of Graphics by Leland Wilkinson.

Grammar of Graphics

A grammar of graphics is a tool that enables us to concisely describe the components of a graphic.

Source: BloggoType

Hello ggplot2!

ggplot2: Elegant Graphics for Data Analysis

Simple example

ggplot(data = mpg,                               # Use the mpg data
       mapping = aes(x = displ, y = hwy)) +      # map (explore) the variables displ and hwy
  geom_point()                                   # visualize the observations with points

colour, size, shape and other aesthetic attributes

ggplot(mpg, aes(displ, hwy, colour = drv))+ 
  geom_point()

ggplot(mpg, aes(displ, hwy, shape = drv))+ 
  geom_point()

ggplot(mpg, aes(displ, hwy, size = drv))+ 
  geom_point()

Faceting

facet_wrap()

ggplot(mpg, aes(displ, hwy)) + 
  geom_point() + 
  facet_wrap(~class)

Other Plot Geoms - geom_smooth()

ggplot(mpg, aes(displ, hwy)) + 
  geom_point() + 
  geom_smooth()

Other Plot Geoms - geom_smooth()

An important argument to geom_smooth() is the method, which allows you to choose which type of model is used to fit the smooth curve.
Default is method = "loess"

ggplot(mpg, aes(displ, hwy)) + 
  geom_point() + 
  geom_smooth(method = "loess")

ggplot(mpg, aes(displ, hwy)) + 
  geom_point() + 
  geom_smooth(method = "lm") #linear model

Other Plot Geoms

ggplot(mpg, aes(drv, hwy))+
  geom_jitter()

ggplot(mpg, aes(drv, hwy))+
  geom_boxplot()

ggplot(mpg, aes(drv, hwy))+
  geom_violin()

Other Plot Geoms

geom_histogram() geom_freqploy()

ggplot(mpg, aes(hwy))+
  geom_histogram()

ggplot(mpg, aes(hwy))+
  geom_freqpoly()

Other Plot Geoms

geom_histogram() geom_freqploy()

Both geoms bin the data. You should experiment with the binwidth or bin arguments when exploring your data.

ggplot(mpg, aes(hwy)) + 
  geom_freqpoly(binwidth = 10)

ggplot(mpg, aes(hwy)) + 
  geom_freqpoly(binwidth = 1)

Other Plot Geoms geom_bar()

ggplot(mpg, aes(x = manufacturer)) + 
  geom_bar()

ggplot(mpg, aes(y =manufacturer)) + 
  geom_bar()

Other Plot Geoms

geom_bar() + fct_infreq()

mpg |> 
ggplot(aes(y = fct_infreq(manufacturer))) + 
  geom_bar()

Other Plot Geoms

geom_bar() + fct_infreq() and aes(fill = [VAR] )

mpg |> 
ggplot(aes(y = fct_infreq(manufacturer), 
           fill = class)) + 
  geom_bar()

Other Plot Geoms

geom_bar() and position = "fill"

position = "fill" works like stacking, but makes each set of stacked bars the same height. This makes it easier to compare proportions across groups.

mpg |> 
ggplot(aes(y = fct_infreq(manufacturer), 
           fill = class)) + 
  geom_bar(position = "fill")

Adding fixedlines

geom_hline() geom_vline() geom_abline()

mpg |> 
  ggplot(aes(x = cty, y = hwy))+
  geom_hline(
    yintercept = 30, 
    color = "blue", 
    size = 2) +
  geom_point()

mpg |> 
  ggplot(aes(x = cty, y = hwy))+
  geom_vline(
    aes(xintercept = mean(cty)), 
    linetype = 2) +
  geom_point()

mpg |> 
  ggplot(aes(x = cty, y = hwy))+
  geom_abline(
    intercept = 0, slope = 1, 
    color = "red", 
    size = 1.5, 
    linetype = 4) +
  geom_point()

labs()- Title, Subtitle, Legends, and more

labs() can do it all

mpg |> 
  ggplot(aes(x = class, y = cty, fill = class))+
  geom_boxplot()+
  labs(
    title = "Milage in cities of various car classes",
    subtitle = "Car classes are colored with nice colors using the fill argument!!",
    x = "Class of car",
    y = "Milage in cities",
    fill = "Class",
    caption = "The data comes from the mpg dataset"
    )

Themes

ggplot comes prepackaged with several different themes

p <- mpg |> 
  ggplot(aes(x = class, y = cty, fill = class))+
  geom_boxplot()

p + theme_bw()

p + theme_classic()

Themes

The package ggthemes have many different themes.

p + ggthemes::theme_stata()
p + ggthemes::theme_economist()
p + ggthemes::theme_excel_new()
p + ggthemes::theme_fivethirtyeight()

Colour and fill scales

Different variables need different colour scales

Colours can represent different kinds of information:

• Sequential - values that go from low to high, e.g. weight or weight classes
• Qualitative - categories with no natural order, e.g. animal species (cat, dog, mouse)
• Diverging - values that move away from a meaningful or neutral midpoint, e.g. change scores where zero means no change

For most plots, the default ggplot colours work well.
When choosing your own scale, viridis is a good general-purpose option. The viridis scale is color blind friendly and prints well on greyscale output.

Categorical data

Here, fill is mapped to a categorical variable.

p <- mpg |> 
  ggplot(aes(x = cty, y = hwy, fill = drv)) +
  geom_point(shape = 21, size = 3)

p

We can use a discrete viridis scale:

p + scale_fill_viridis_d()

Continuous data

Here, colour is mapped to a continuous variable.

p <- mpg |> 
  ggplot(aes(x = cty, y = hwy, colour = displ)) +
  geom_point(size = 3)

p

Use a continuous viridis scale:

p +  scale_colour_viridis_c()

Use a continuous viridis scale, but another option:

p +  scale_colour_viridis_c(option = "magma")

Diverging scales

For diverging data its often a good idea to create your own scale.

For this example we create some sample data

dplyr recap - What does this code do?

# Create some sample data
sample_data <- 
  tibble(
    change = rnorm(100, mean = 0, sd = 10), 
    group = rep(c("A", "B"), each = 50)) |> 
  mutate(
    change = if_else(group == "B",
                     change + rnorm(n = 50, 
                                    mean = 5, 
                                    sd = 5),
                     change)
    ) 

sample_data

# A tibble: 100 × 2
    change group
     <dbl> <chr>
 1  -8.77  A    
 2  -2.10  A    
 3 -17.4   A    
 4  -3.29  A    
 5   2.38  A    
 6   7.81  A    
 7  -7.90  A    
 8  -8.39  A    
 9  -6.56  A    
10  -0.975 A    
# ℹ 90 more rows

Diverging scales

p <- sample_data |>  
  ggplot(aes (x = group, y = change, color = change))+
  geom_hline(yintercept = 0)+
  geom_jitter(width = 0.2, height = 0, size = 3) 

p

p +
  scale_colour_gradient2(
   low = "blue",
   mid = "white",
   high = "red",
   midpoint = 0
  )

The patchwork package

Arranging plots with the patchwork package

p1 <- ggplot(mpg) + 
  geom_point(aes(x = displ, y = hwy))+
  ggtitle("p1")

p2 <- ggplot(mpg) + 
  geom_bar(aes(x = as.character(year), fill = drv), position = "dodge") + 
  labs(x = "year")+
  ggtitle("p2")

p3 <- ggplot(mpg) + 
  geom_histogram(aes(x = hwy, fill = drv), colour = NA) + 
  facet_grid(rows = vars(drv))+
  ggtitle("p3")

p4 <- ggplot(mpg) + 
  stat_summary(aes(x = drv, y = hwy, fill = drv), geom = "col", fun.data = mean_se) +
  stat_summary(aes(x = drv, y = hwy), geom = "errorbar", fun.data = mean_se, width = 0.5)+
  ggtitle("p4")

Arranging plots with the patchwork package

Arranging plots with the patchwork package

• + does not specify any specific layout, only that the plots should be displayed together.

• patchwork automagically decides the layout.

• Adding 3 plots together will create a 1x3 grid.

• Adding 4 plots together will create a 2x2 grid.

p1 + p2

Arranging plots with the patchwork package

• + does not specify any specific layout, only that the plots should be displayed together.

• patchwork automagically decides the layout.

• Adding 3 plots together will create a 1x3 grid.

• Adding 4 plots together will create a 2x2 grid.

p1 + p2 + p3 +p4

Arranging plots with the patchwork package

plot_layout()

You have a very high degree of control via the plot_layout() function

p1 + p2 + p3 + p4 + plot_layout(nrow = 1)

Arranging plots with the patchwork package

/ and |

/ forces a single column.
| forces a single row.

p1/p2

p1|p2 # sames as p1 + p2

Arranging plots with the patchwork package

nesting plots

p3 | (p2 /(p1 | p4))

Arranging plots with the patchwork package

guides = "collect"

p1 + p2 + p3  + plot_layout(ncol = 2, guides = "collect")

Arranging plots with the patchwork package

guide_area()

p1 + p2 + p3 + plot_layout(ncol = 2, guides = "collect") + guide_area() 

Arranging plots with the patchwork package

patchwork comes with the powerful & operator that allows you change the settings on all the individual plots.

p1 + p2 + p3 + 
  plot_layout(ncol = 2, guides = "collect") + 
  guide_area() +
  scale_fill_brewer(type = "qual",
                    palette = 3) +
  theme_classic()

Can you spot the &s in the code below?

p1 + p2 + p3 + 
  plot_layout(ncol = 2, guides = "collect") + 
  guide_area() &   
  scale_fill_brewer(type = "qual",
                    palette = 3) &
  theme_classic()