Multiple plots in one figure

Data Visualisation with R

👩🏻‍💻 Emi Tanaka @ Monash University

• emi.tanaka@monash.edu
• @statsgen
• github.com/emitanaka
• emitanaka.org

28th November 2022 Australasian Applied Statistics Conference 2022

Facets

• Faceting in ggplot2 refers to partitioning of the plot region to multiple panels each showing a different subset of the data.

• There are three types of faceting:

  • facet_null() (the default),
  • facet_wrap(), and
  • facet_grid().

• We don’t really use facet_null() much.

Illustrative data 🌱 growth of soybean

data(davidian.soybean, package = "agridat")
str(davidian.soybean)

'data.frame':   412 obs. of  5 variables:
 $ plot   : Factor w/ 48 levels "1988F1","1988F2",..: 1 1 1 1 1 1 1 1 1 1 ...
 $ variety: Factor w/ 2 levels "F","P": 1 1 1 1 1 1 1 1 1 1 ...
 $ year   : int  1988 1988 1988 1988 1988 1988 1988 1988 1988 1988 ...
 $ day    : int  14 21 28 35 42 49 56 63 70 77 ...
 $ weight : num  0.106 0.261 0.666 2.11 3.56 ...

• The data contains:
  • plot: the plot code
  • variety: the code (F = commercial and P = experimental) for which soybean variety
  • year: the year it was grown (1988-1990)
  • day days after planting
  • weight weight of soybean leaves

facet_wrap()

• facet_wrap() divides the data into mutually exclusive subset based on the input variable and create subplots based on the data in the corresponding panel.

ggplot(davidian.soybean, aes(day, weight)) + 
  geom_point() +
  facet_wrap(~year)

Wrapping many panels

• facet_wrap “wraps” the panels into 2d if you have many panels

ggplot(davidian.soybean, aes(day, weight)) + 
  geom_point() +
  facet_wrap(~plot)

Wrapping by a continuous variable

• Discretize continuous variable using cut(), cut_width(), cut_interval() or cut_number().

ggplot(davidian.soybean, aes(day, weight)) + 
  geom_point() +
  facet_wrap(~cut_number(day, 4))

Changing the scales for each panel

• By default the scales are shared across the panels but you can choose to have scales built independently for each panel:

ggplot(davidian.soybean, 
       aes(day, weight)) + 
  geom_point() +
  facet_wrap(~cut_number(day, 4), 
             scales = "free")

• If you want the scale to be independent only in one dimension then you can use scales = "free_x" and scales = "free_y".

Wrapping with more than one variables

• If you want to wrap the facet by more than one variable, you can combine the variables by using +:

ggplot(davidian.soybean, 
       aes(day, weight)) + 
  geom_point() +
  facet_wrap(~ year + variety)

• Alternatively, you can use vars() instead of +.

facet_grid()

• Alternatively, you can use facet_grid() to lay out the panels in a 2d grid:

ggplot(davidian.soybean, aes(day, weight)) + 
  geom_point() +
  facet_grid(variety ~ year)

Missing facet variables

• If a layer data is missing the faceting variable(s), then the layer will be shown across all panels based on all of the layer data

ggplot(davidian.soybean, aes(day, weight)) + 
  geom_point(data = function(.data) select(.data, -c(variety, year)), color = "grey70") +
  geom_point() +
  facet_grid(variety ~ year)

• This is a nice way to creating a “shadow” across all panels.

Combining plots with patchwork

• We use the patchwork package for combining plots.

• Combining multiple ggplot objects is then just a matter of using + (side-by-side) or / (top-to-bottom).

Demo with patchwork

library(patchwork)
g1 <- ggplot(davidian.soybean, aes(day, weight)) +
  geom_point(aes(color = as.factor(year))) 

g2 <- ggplot(davidian.soybean, aes(as.factor(year), weight)) +
  geom_jitter(aes(color = as.factor(year))) 

g1 + g2

Combining plot layouts

• patchwork can collect the same guides across the plots as well as automatically add tag labels

g1 + g2 + 
  plot_layout(guides = "collect") + 
  plot_annotation(tag_levels = "A")

Plot layout

• You can combine the use of + and / to layout plots horizontally or vertically using () to group rows or columns

(g1 + g2) / (g2 + g1) + 
  plot_layout(guides = "collect")

Flexible plot layout

• If you want even more control over the plot layout, then you can specify a string with the structure you want like below:

design <- "
 11###
 11###
 #2222"

g1 + g2 + 
  plot_layout(guides = "collect",
              design = design)

Time to code!

15:00

> Go to emitanaka.org/dataviz-workshop/exercises/
> Click Exercise 6