Parts of a Whole

Parts of a Whole

Abstract

Slices, Portions, Counts, and Aggregates of Data

“There is no such thing as a”self-made” man. We are made up of thousands of others. Everyone who has ever done a kind deed for us, or spoken one word of encouragement to us, has entered into the make-up of our character and of our thoughts.”

— George Matthew Adams, newspaper columnist (23 Aug 1878-1962)

Setting up the Packages

library(tidyverse)
library(mosaic)
library(ggformula)
library(plotrix) # Fan, Pyramid Chart
# devtools::install_github("zmeers/ggparliament")
library(ggparliament) # Parliament Chart
library(ggpol) # Parliament, Arc-Bar and other interesting charts
library(data.tree) # Many plots related to heirarchical data
# install.packages("waffle", repos = "https://cinc.rud.is")
library(waffle)
library(tidygraph) # Trees, Dendros, and Circle Packings
library(ggraph) # Trees, Dendros, and Circle Packings
library(echarts4r) # Interactive Charts

library(patchwork) # Arrange your plots

##
library(tidyplots) # Easily Produced Publication-Ready Plots
library(tinyplot) # Plots with Base R
library(tinytable) # Elegant Tables for our data

## ggplot theme
library(hrbrthemes)
hrbrthemes::import_roboto_condensed() # Import Roboto Condensed font for use in charts
hrbrthemes::update_geom_font_defaults() # Update matching font defaults for text geoms
ggplot2::theme_set(new = theme_classic(base_family = "Roboto Condensed")) # Set consistent graph theme

What Graphs will we see today?

There are a good few charts available to depict things that constitute other bigger things. We will discuss a few of these: Pie, Fan, and Donuts; Waffle and Parliament charts; Trees, Dendrograms, and Circle Packings. (The last three visuals we will explore along with network diagrams in a later module.)

Pies and Fans

So let us start with “eating humble pie”: discussing a Pie chart first.

A pie chart is a circle divided into sectors that each represent a proportion of the whole. It is often used to show percentage, where the sum of the sectors equals 100%.

The problem is that humans are pretty bad at reading angles. This ubiquitous chart is much vilified in the industry and bar charts that we have seen earlier, are viewed as better options. On the other hand, pie charts are ubiquitous in business circles, and are very much accepted! Do also read this spirited defense of pie charts here. https://speakingppt.com/why-tufte-is-flat-out-wrong-about-pie-charts/

And we will also see that there is an attractive, and similar-looking alternative, called a fan chart which we will explore here.

Using Base R

Base R has a simple pie command that does the job. Let’s create some toy data first:

pie_data <- tibble(
  sales = c(0.12, 0.3, 0.26, 0.16, 0.04, 0.12),

  # Labels MUST be character entries for `pie` to work
  labels = c(
    "Blueberry", "Cherry", "Apple", "Boston Cream",
    "Other", "Vanilla Cream"
  )
)
pie_data
pie(
  x = pie_data$sales,
  labels = pie_data$labels, # Character Vector is a MUST

  # Pie is within a square of 1 X 1 units
  # Reduce radius if needed to see labels properly
  radius = 0.95,
  init.angle = 90, # First slice starts at 12 o'clock position

  # Change the default colours. Comment this and see what happens.
  col = grDevices::hcl.colors(palette = "Plasma", n = 6)
)

ABCDEFGHIJ0123456789

sales <dbl>	labels <chr>
0.12	Blueberry
0.30	Cherry
0.26	Apple
0.16	Boston Cream
0.04	Other
0.12	Vanilla Cream

6 rows

Using ggformula

We create a bar chart or a column chart as appropriate, with bars filled by category. The width parameter is set to 1 so that the bars touch. The bars have a fixed width along the x-axis; the height of the bar varies based on the number we wish to show. Then the coord_polar(theta = "y") converts the bar plot into a pie.

ggplot2::theme_set(new = theme_classic(base_family = "Roboto Condensed")) # Set consistent graph theme

# Using gf_col since we have a count/value column already
pie_data %>%
  gf_col(sales ~ 1, fill = ~labels, width = 1, color = "black") %>%
  gf_refine(scale_fill_brewer(palette = "Set1"))
pie_data %>%
  gf_col(sales ~ 1, fill = ~labels, width = 1, color = "black") %>%
  gf_refine(coord_polar(theta = "y")) %>%
  gf_refine(scale_fill_brewer(palette = "Set1"))
# Using gf_bar since we don't have ready made counts
gf_bar(
  data = mpg,
  ~1,
  fill = ~drv,
  color = "black", # border for the bars/slices
  width = 1
) %>%
  gf_refine(scale_fill_brewer(palette = "Set1"))
gf_bar(
  data = mpg,
  ~0.5,
  fill = ~drv,
  color = "black", # border for the bars/slices
  width = 1
) %>%
  gf_refine(scale_fill_brewer(palette = "Set1")) %>%
  gf_theme(theme_minimal()) %>%
  gf_theme(theme(
    axis.line.y = element_blank(),
    axis.text.y = element_blank(),
    axis.title.y = element_blank()
  )) %>%
  gf_refine(coord_polar(theta = "y"))

Using echarts4r

Here is a basic interactive pie chart withecharts4r:

pie_data <- tibble(
  sales = c(0.12, 0.3, 0.26, 0.16, 0.04, 0.12),
  labels = c(
    "Blueberry", "Cherry", "Apple", "Boston Cream", "Other",
    "Vanilla Cream"
  )
)
pie_data %>%
  e_charts(x = labels) %>%
  e_pie(
    serie = sales, clockwise = TRUE,
    startAngle = 90
  ) %>%
  e_legend(list(
    orient = "vertical",
    left = "right"
  )) %>%
  e_tooltip()

We can add more bells and whistles to the humble-pie chart, and make a Nightingale rosechart out of it:

pie_data <- tibble(
  sales = c(0.12, 0.3, 0.26, 0.16, 0.04, 0.12),
  labels = c(
    "Blueberry", "Cherry", "Apple", "Boston Cream", "Other",
    "Vanilla Cream"
  )
)
pie_data %>%
  e_charts(x = labels) %>%
  e_pie(
    serie = sales, clockwise = TRUE,
    startAngle = 90,
    roseType = "area"
  ) %>% # try "radius"

  # Lets move the legend
  e_legend(left = "right", orient = "vertical") %>%
  e_tooltip()
pie_data %>%
  e_charts(x = labels) %>%
  e_pie(
    serie = sales, clockwise = TRUE,
    startAngle = 90,
    roseType = "radius"
  ) %>%
  # Lets move the legend
  e_legend(left = "right", orient = "vertical") %>%
  e_tooltip()

For more information and customization look at https://echarts.apache.org/en/option.html#series-pie

The fan Plot

The fan plot (from the plotrix package) displays numerical values as arcs of overlapping sectors. This allows for more effective comparison:

ggplot2::theme_set(new = theme_classic(base_family = "Roboto Condensed")) # Set consistent graph theme

plotrix::fan.plot(
  x = pie_data$sales,
  labels = pie_data$labels,
  col = grDevices::hcl.colors(palette = "Lajolla", n = 6), # Try hcl.pals()
  shrink = 0.03,
  # How much to shrink each successive sector

  label.radius = 1.15,
  main = "Fan Plot of Ice Cream Flavours",
  # ticks = 360,
  # if we want tick marks on the circumference

  max.span = pi
)

There is no fan plot possible with echarts4r, as far as I know.

The Donut Chart

The donut chart suffers from the same defects as the pie, so should be used with discretion. The donut chart is essentially a gf_rect from ggformula, plotted on a polar coordinate set of of axes:

Using ggformula

Let us make some toy data:

ggplot2::theme_set(new = theme_classic(base_family = "Roboto Condensed")) # Set consistent graph theme

# Data
df <- tibble(
  group = LETTERS[1:3],
  value = c(25, 20, 35)
)

df <-
  df %>%
  dplyr::mutate(
    fraction = value / sum(value), # percentages
    ymax = cumsum(fraction), # cumulative percentages
    ymin = lag(ymax, 1, default = 0),
    # bottom edge of each
    label = paste0(group, "\n value: ", value),
    labelPosition = (ymax + ymin) / 2 # labels midway on arcs
  )

df
df %>%
  # gf_rect() formula: ymin + ymax ~ xmin + xmax
  # Bars with varying thickness (y) proportional to data
  # Fixed length x (2 to 4)
  gf_rect(ymin + ymax ~ 2 + 4,
    fill = ~group, colour = "black"
  ) %>%
  gf_label(labelPosition ~ 3.5,
    label = ~label,
    size = 4
  ) %>%
  # When switching to polar coords:
  # x maps to radius
  # y maps to angle theta
  # so we create a "hole" in the radius, in x
  gf_refine(coord_polar(
    theta = "y",
    direction = 1
  )) %>%
  # Up to here will give us a pie chart

  # Now to create the hole
  # try to play with the "0"
  # Recall x = [2,4]
  gf_refine(xlim(c(-2, 5)), scale_fill_brewer(palette = "Spectral")) %>%
  gf_theme(theme = theme_void()) %>%
  gf_theme(legend.position = "none")

ABCDEFGHIJ0123456789

group <chr>	value <dbl>	fraction <dbl>	ymax <dbl>	ymin <dbl>	label <chr>
A	25	0.3125	0.3125	0.0000	A\n value: 25
B	20	0.2500	0.5625	0.3125	B\n value: 20
C	35	0.4375	1.0000	0.5625	C\n value: 35

3 rows | 1-6 of 7 columns

Using echarts4r

The donut chart is simply a variant of the pie chart in echarts4r:

df <- tibble(
  group = LETTERS[1:3],
  value = c(25, 20, 35)
)

df <-
  df %>%
  dplyr::mutate(
    fraction = value / sum(value), # percentages
    ymax = cumsum(fraction), # cumulative percentages
    ymin = lag(ymax, 1, default = 0),
    # bottom edge of each
    label = paste0(group, "\n value: ", value),
    labelPosition = (ymax + ymin) / 2 # labels midway on arcs
  )
df
df %>%
  e_charts(x = group, width = 400) %>%
  e_pie(
    serie = value,
    clockwise = TRUE,
    startAngle = 90,
    radius = c("50%", "70%")
  ) %>%
  e_legend(left = "right", orient = "vertical") %>%
  e_tooltip()

ABCDEFGHIJ0123456789

group <chr>	value <dbl>	fraction <dbl>	ymax <dbl>	ymin <dbl>
A	25	0.3125	0.3125	0.0000
B	20	0.2500	0.5625	0.3125
C	35	0.4375	1.0000	0.5625

3 rows | 1-5 of 7 columns

Waffle Charts

Waffle charts are often called “square pie charts” !

Here we will need to step outside of ggformula and get into ggplot itself momentarily. (Always remember that ggformula is a simplified and intuitive method that runs on top of ggplot.) We will use the waffle package.

# install.packages("waffle", repos = "https://cinc.rud.is")
library(waffle)

# Data
df <- tibble(
  group = LETTERS[1:3],
  value = c(25, 20, 35)
)
df

ABCDEFGHIJ0123456789

group <chr>	value <dbl>
A	25
B	20
C	35

3 rows

# Waffle plot
# Using ggplot, sadly not yet ggformula

ggplot2::theme_set(new = theme_classic(base_family = "Roboto Condensed")) # Set consistent graph theme

ggplot(df, aes(fill = group, values = value)) +
  geom_waffle(
    n_rows = 8,
    size = 0.33,
    colour = "white",
    na.rm = TRUE
  ) +
  scale_fill_manual(
    name = NULL,
    values = c("#BA182A", "#FF8288", "#FFDBDD"),
    labels = c("A", "B", "C")
  ) +
  coord_equal() +
  theme_void()

Parliament Charts

The package ggpol offers an interesting visualization in the shape of a array of “seats” in a parliament. (There is also a package called ggparliament which in my opinion is a bit cumbersome, having a two-step procedure to convert data into “parliament form” etc. )

ggplot2::theme_set(new = theme_classic(base_family = "Roboto Condensed")) # Set consistent graph theme

# Same toy dataset
# df <- tibble(group = LETTERS[1:3],
#                  value = c(25, 20, 35))
#
# Parliament Plot
ggplot(df) +
  ggpol::geom_parliament(
    aes(
      seats = value,
      fill = group
    ),
    r0 = 2, # inner radius
    r1 = 4 # Outer radius
  ) +
  scale_fill_manual(
    name = NULL,
    values = c("#BA182A", "#FF8288", "#FFDBDD"),
    labels = c("A", "B", "C")
  ) +
  coord_equal() +
  theme_void()

Trees, Dendrograms, and Circle Packings

There are still more esoteric plots to explore, if you are hell-bent on startling people ! There is an R package called ggraph, that can do these charts, and many more:

ggraph is an extension of ggplot2 aimed at supporting relational data structures such as networks, graphs, and trees. While it builds upon the foundation of ggplot2 and its API it comes with its own self-contained set of geoms, facets, etc., as well as adding the concept of layouts to the grammar.

We will explore these charts when we examine network diagrams. For now, we can quickly see what these diagrams look like. Although the R-code is visible to you, it may not make sense at the moment!

Dendrograms

From the R Graph Gallery Website :

Dendrograms can be built from:

• Hierarchical dataset: think about a CEO managing team leads managing employees and so on.

• Clustering result: clustering divides a set of individuals in group according to their similarity. Its result can be visualized as a tree.

# create an edge list data frame giving the hierarchical structure of your individuals
d1 <- tibble(from = "origin", to = paste("group", seq(1, 5), sep = ""))
d2 <- tibble(from = rep(d1$to, each = 5), to = paste("subgroup", seq(1, 25), sep = "_"))
edges <- rbind(d1, d2)
edges

ABCDEFGHIJ0123456789

from <chr>	to <chr>
origin	group1
origin	group2
origin	group3
origin	group4
origin	group5
group1	subgroup_1
group1	subgroup_2
group1	subgroup_3
group1	subgroup_4
group1	subgroup_5

Next

123

Previous

1-10 of 30 rows

# Create a graph object
mygraph1 <- tidygraph::as_tbl_graph(edges)
mygraph1

# A tbl_graph: 31 nodes and 30 edges
#
# A rooted tree
#
# Node Data: 31 × 1 (active)
   name      
   <chr>     
 1 origin    
 2 group1    
 3 group2    
 4 group3    
 5 group4    
 6 group5    
 7 subgroup_1
 8 subgroup_2
 9 subgroup_3
10 subgroup_4
# ℹ 21 more rows
#
# Edge Data: 30 × 2
   from    to
  <int> <int>
1     1     2
2     1     3
3     1     4
# ℹ 27 more rows

ggplot2::theme_set(new = theme_classic(base_family = "Roboto Condensed")) # Set consistent graph theme
# Basic tree
ggraph(mygraph1,
  layout = "dendrogram",
  circular = TRUE
) +
  geom_edge_diagonal() +
  geom_node_point(size = 3) +
  geom_node_label(aes(label = name),
    size = 3, repel = TRUE
  ) +
  theme(aspect.ratio = 1)

# create a data frame
data <- tibble(
  level1 = "CEO",
  level2 = c(rep("boss1", 4), rep("boss2", 4)),
  level3 = paste0("mister_", letters[1:8])
)

# transform it to a edge list!
edges_level1_2 <- data %>%
  select(level1, level2) %>%
  unique() %>%
  rename(from = level1, to = level2)

edges_level2_3 <- data %>%
  select(level2, level3) %>%
  unique() %>%
  rename(from = level2, to = level3)

edge_list <- rbind(edges_level1_2, edges_level2_3)
edge_list

ABCDEFGHIJ0123456789

from <chr>	to <chr>
CEO	boss1
CEO	boss2
boss1	mister_a
boss1	mister_b
boss1	mister_c
boss1	mister_d
boss2	mister_e
boss2	mister_f
boss2	mister_g
boss2	mister_h

1-10 of 10 rows

mygraph2 <- as_tbl_graph(edge_list)
mygraph2

# A tbl_graph: 11 nodes and 10 edges
#
# A rooted tree
#
# Node Data: 11 × 1 (active)
   name    
   <chr>   
 1 CEO     
 2 boss1   
 3 boss2   
 4 mister_a
 5 mister_b
 6 mister_c
 7 mister_d
 8 mister_e
 9 mister_f
10 mister_g
11 mister_h
#
# Edge Data: 10 × 2
   from    to
  <int> <int>
1     1     2
2     1     3
3     2     4
# ℹ 7 more rows

ggplot2::theme_set(new = theme_classic(base_family = "Roboto Condensed")) # Set consistent graph theme

# Now we can plot that
ggraph(mygraph2, layout = "dendrogram", circular = FALSE) +
  geom_edge_diagonal() +
  geom_node_point(size = 3) +
  geom_node_label(aes(label = name), repel = TRUE) +
  theme_void()

Circle Packing

graph_flare <- tbl_graph(flare$vertices, flare$edges)
graph_flare

# A tbl_graph: 252 nodes and 251 edges
#
# A rooted tree
#
# Node Data: 252 × 3 (active)
   name                                            size shortName            
   <chr>                                          <dbl> <chr>                
 1 flare.analytics.cluster.AgglomerativeCluster    3938 AgglomerativeCluster 
 2 flare.analytics.cluster.CommunityStructure      3812 CommunityStructure   
 3 flare.analytics.cluster.HierarchicalCluster     6714 HierarchicalCluster  
 4 flare.analytics.cluster.MergeEdge                743 MergeEdge            
 5 flare.analytics.graph.BetweennessCentrality     3534 BetweennessCentrality
 6 flare.analytics.graph.LinkDistance              5731 LinkDistance         
 7 flare.analytics.graph.MaxFlowMinCut             7840 MaxFlowMinCut        
 8 flare.analytics.graph.ShortestPaths             5914 ShortestPaths        
 9 flare.analytics.graph.SpanningTree              3416 SpanningTree         
10 flare.analytics.optimization.AspectRatioBanker  7074 AspectRatioBanker    
# ℹ 242 more rows
#
# Edge Data: 251 × 2
   from    to
  <int> <int>
1   221     1
2   221     2
3   221     3
# ℹ 248 more rows

ggplot2::theme_set(new = theme_classic(base_family = "Roboto Condensed")) # Set consistent graph theme

set.seed(1)
ggraph(graph_flare, "circlepack", weight = size) +
  geom_node_circle(aes(fill = as_factor(depth)), size = 0.25, n = 50) +
  coord_fixed() +
  scale_fill_brewer(name = "Depth", palette = "Set1")

Your Turn

1. Use the penguins dataset from the palmerpenguins package and plot pies, fans, and donuts as appropriate.
2. Look at the whigs and highschool datasets in the package ggraph. Plot Pies, Fans and if you are feeling confident, Trees, Dendrograms, and Circle Packings as appropriate for these.

References

1. Iaroslava.2020. A Parliament Diagram in R, https://datavizstory.com/a-parliament-diagram-in-r/
   
2. Venn Diagrams in R, Venn diagram in ggplot2 | R CHARTS (r-charts.com)
   
3. Generate icon-array charts without code! https://iconarray.com

R Package Citations

Package	Version	Citation
data.tree	1.1.0	Glur (2023)
echarts4r	0.4.5	Coene (2023)
ggparliament	3.1.6	Hickman, Meers, and Leeper (2024)
ggpol	0.0.7	Tiedemann (2020)
ggraph	2.2.1	Pedersen (2024a)
plotrix	3.8.4	J (2006)
tidygraph	1.3.1	Pedersen (2024b)
waffle	1.0.2	Rudis and Gandy (2023)

Coene, John. 2023. Echarts4r: Create Interactive Graphs with “Echarts JavaScript” Version 5. https://doi.org/10.32614/CRAN.package.echarts4r.

Glur, Christoph. 2023. data.tree: General Purpose Hierarchical Data Structure. https://doi.org/10.32614/CRAN.package.data.tree.

Hickman, Robert, Zoe Meers, and Thomas J. Leeper. 2024. ggparliament: Parliament Plots. https://github.com/zmeers/ggparliament.

J, Lemon. 2006. “Plotrix: A Package in the Red Light District of r.” R-News 6 (4): 8–12.

Pedersen, Thomas Lin. 2024a. ggraph: An Implementation of Grammar of Graphics for Graphs and Networks. https://doi.org/10.32614/CRAN.package.ggraph.

———. 2024b. tidygraph: A Tidy API for Graph Manipulation. https://doi.org/10.32614/CRAN.package.tidygraph.

Rudis, Bob, and Dave Gandy. 2023. waffle: Create Waffle Chart Visualizations. https://doi.org/10.32614/CRAN.package.waffle.

Tiedemann, Frederik. 2020. ggpol: Visualizing Social Science Data with “ggplot2”. https://doi.org/10.32614/CRAN.package.ggpol.