Analyzing Unicorn Companies with treemaps in R

Have you ever heard about Unicorn companies? This post shows a list of Unicorn companies around the world, analyzing them with data analysis techniques, using the R language.

A Unicorn company is usually defined as a private startup valued at over U$ 1 billion.

The full and updated list of Unicorn companies in the world used on this analysis can be found here, at the CBInsights website.

Note: the data for this post was downloaded on 1st October 2022.

What will you see in this blog post?

• The countries and industries with the most valuable Unicorn companies.
• Treemaps that will allow you to explore all the Unicorn companies, split into quartiles.
• A treemap showing the Unicorn companies in Brazil (my country) :)

What technical skills will you find here?

• How to deal with data analysis in a real world, using R language.
• Properly transform the data before plotting it.
• Use some good practices of data visualization.

In this kind of blog post, I won’t write any interpretations about the charts and tables, because I want to be able to update the data in the future without needing to replace text paragraphs. So, the idea is to keep the charts as self-explanatory as possible.

Reading data: Unicorn companies list

First, we need to load the used packages and read the data. Below I am showing the first rows of the raw data and doing some data manipulations.

In this very beginning of the post some settings are already created, like the colors palette for Countries and Industries, to use in the whole document.

The next chunk of code is responsible to load the packages, read the data and create some useful variables, like valuation Quartiles, Number of Investors and Years as Unicorn.

Code

# Loading packages
library(dplyr)
library(echarts4r)
library(lubridate)
library(MetBrewer)
library(reactable)
library(readxl)

# Reading the Excel file
unicorn_data <- readxl::read_excel("unicorn_companies.xlsx", skip = 2) |>
  dplyr::arrange(dplyr::desc("Valuation ($B)")) |>
  dplyr::mutate(
    `Years as Unicorn` = round(
      (lubridate::interval(`Date Joined`, lubridate::today()) %/% months(1)) / 12,
      digits = 2
    ),
    `Number of Investors` = stringr::str_count(`Select Investors`, ',') + 1,
    `Industry` = as.factor(`Industry`),
    Quartiles = dplyr::case_when(
      `Valuation ($B)` <= quantile(`Valuation ($B)`, 0.25, na.rm = TRUE) ~ 'Q4',
      `Valuation ($B)` <= quantile(`Valuation ($B)`, 0.5, na.rm = TRUE) ~ 'Q3',
      `Valuation ($B)` <= quantile(`Valuation ($B)`, 0.75, na.rm = TRUE) ~ 'Q2',
      `Valuation ($B)` <= quantile(`Valuation ($B)`, 1, na.rm = TRUE) ~ 'Q1'
    )
  )

Valuation by Country and Industry

After reading the data I have created two charts to briefly show the summarized data by Country and by Industry.

In this part I also create the color palettes that will be used in other charts as well.

Here comes a nice tip: the usage of the R package MetBrewer to create very nice color palettes.

Code

create_color_palette <- function(vector, theme) {
  dplyr::tibble(
    name = vector,
    itemStyle = dplyr::tibble(
      color = as.character(MetBrewer::met.brewer(theme, length(vector)))
    )
  )
}

unicorn_data_by_country <- unicorn_data |>
  dplyr::group_by(Country) |>
  dplyr::summarise(`Valuation ($B)` = sum(`Valuation ($B)`)) |>
  dplyr::arrange(dplyr::desc(`Valuation ($B)`)) |>
  head(15)

colors_country <- create_color_palette(unicorn_data_by_country$Country, "Derain")

unicorn_data_by_industry <- unicorn_data |>
  dplyr::group_by(Industry) |>
  dplyr::summarise(`Valuation ($B)` = sum(`Valuation ($B)`)) |>
  dplyr::arrange(dplyr::desc(`Valuation ($B)`))

colors_industry <- create_color_palette(unicorn_data_by_industry$Industry, "Juarez")

Below we can see the sum of all the companies’ Valuations for Industry and Country in a column chart.

Code

unicorn_data_by_industry |>
  dplyr::left_join(
    colors_industry |> dplyr::rename(Industry = name, color = `itemStyle`),
    by = "Industry"
  ) |>
  echarts4r::e_charts(`Valuation ($B)`) |>
  echarts4r::e_legend(show = FALSE) |>
  echarts4r::e_bar(Industry) |>
  echarts4r::e_y_axis(type = "category") |>
  echarts4r::e_grid(containLabel = TRUE) |>
  echarts4r::e_tooltip() |>
  echarts4r::e_add_nested('itemStyle', color) |>
  echarts4r::e_title(text = "Unicorns by Industry", subtext = "Sum the valuation of all the Industries")

Code

unicorn_data_by_country |>
  dplyr::left_join(
    colors_country |> dplyr::rename(Country = name, color = `itemStyle`),
    by = "Country"
  ) |> 
  echarts4r::e_charts(`Valuation ($B)`) |>
  echarts4r::e_legend(show = FALSE) |>
  echarts4r::e_bar(Country) |>
  echarts4r::e_y_axis(type = "category") |>
  echarts4r::e_grid(containLabel = TRUE) |>
  echarts4r::e_tooltip() |>
  echarts4r::e_add_nested('itemStyle', color) |>
  echarts4r::e_title(text = "Unicorns by Country", subtext = "Sum the valuation of the 15 first countries") 

Exploring the top 50 Unicorn companies

In this section I filtered the data just to show the first 50 Unicorn companies with the highest valuation.

We will see here an interactive table and also two treemaps, showing the data by Country and by Industry.

I won’t show all the table here, because if you want to have access to the complete dataset, please check the CBInsights website.

In the table it is possible to sort the data by column and also search.

Code

interactive_table_data <- unicorn_data |>
  dplyr::select(-c("Date Joined", "Number of Investors", "Quartiles")) |>
  dplyr::mutate(`Years as Unicorn` = floor(`Years as Unicorn`)) |>
  head(50)

# Creating an interactive table
reactable::reactable(
  interactive_table_data,
  columns = list(
    `Select Investors` = colDef(minWidth = 180),
    Country = colDef(html = TRUE)
  ),
  style = list(fontSize = "0.70rem"),
  defaultPageSize = 5,
  searchable = TRUE,
  striped = TRUE,
  resizable = TRUE
)

Below it is possible to check two treemaps built with the R language. In the next section all the companies will be visible, but here we can see only the first 50.

The idea to show the following two treemaps is that as the number of companies is not so big, we can show all the 50 companies in only one view. So the user can compare companies that belong to different categories in the same chart.

If you want to explore all the companies from the data, please check the next section.

Code

treemap_tooltip <- htmlwidgets::JS(
      "function(info){
        var treePathInfo = info.treePathInfo;
        var treePath = [];
        for (var i = 1; i < treePathInfo.length; i++) {
          treePath.push(treePathInfo[i].name);
        }
      return(
        '<strong>' + echarts.format.encodeHTML(treePath.join(' / ')) + '</strong>' + '<br>' +
        'Valuation ($B): <i>' + info.value + '</i><br>'
      )}"
      )

unicorn_data |>
  dplyr::arrange(dplyr::desc(`Valuation ($B)`)) |>
  head(50) |>
  dplyr::select(Country, name = Company, value = `Valuation ($B)`) |>
  tidyr::nest(children = c("name", "value")) |>
  dplyr::mutate(value = purrr::map_dbl(children, ~ sum(.x$value))) |>
  dplyr::rename(name = Country) |>
  dplyr::left_join(colors_country, by = "name") |>
  echarts4r::e_charts() |>
  echarts4r::e_title(text = "Unicorns by Country") |>
  echarts4r::e_treemap(
    leafDepth = 2,
    itemStyle = list(normal = list(
      borderWidth = 0,
      gapWidth = 2,
      backgroundColor = "white"
    )),
    upperLabel = list(
      normal = list(
        show = TRUE,
        height = 30,
        formatter = "{b}",
        color = "black",
        fontSize = 12
      )
    )
  ) |>
  echarts4r::e_tooltip(
    backgroundColor = "rgba(255,255,255,0.8)",
    formatter = treemap_tooltip
  )

Code

unicorn_data |>
  dplyr::arrange(dplyr::desc(`Valuation ($B)`)) |>
  head(50) |>
  dplyr::select(Industry, name = Company, value = `Valuation ($B)`) |>
  tidyr::nest(children = c("name", "value")) |>
  dplyr::mutate(value = purrr::map_dbl(children, ~ sum(.x$value))) |>
  dplyr::rename(name = Industry) |>
  dplyr::left_join(colors_industry, by = "name") |>
  echarts4r::e_charts() |>
  echarts4r::e_title(text = "Unicorns by Industry") |>
  echarts4r::e_treemap(
    leafDepth = 2,
    itemStyle = list(normal = list(
      borderWidth = 0,
      gapWidth = 2,
      backgroundColor = "white"
    )),
    upperLabel = list(
      normal = list(
        show = TRUE,
        height = 30,
        formatter = "{b}",
        color = "black",
        fontSize = 12
      )
    )
  ) |>
  echarts4r::e_tooltip(
    backgroundColor = "rgba(255,255,255,0.8)",
    formatter = treemap_tooltip
  )

Exploring all Unicorn companies with treemaps

This section will show treemaps that allow the user to explore all the Unicorn companies in the dataset. For that, I created a function for the treemap(next chunk of code) and I also decided to plot them by quartiles of valuation, to facilitate the navigation.

Code

tree_map_chart <- function(data, quartile, colors_df, tooltip = treemap_tooltip) {
  filtered_data <- data |> dplyr::filter(Quartiles == quartile) 

  treemap_data <- filtered_data |>
    dplyr::arrange(dplyr::desc(`Valuation ($B)`)) |>
    dplyr::filter(Quartiles == quartile) |>
    dplyr::select(Industry, name = Company, value = `Valuation ($B)`) |>
    tidyr::nest(children = c("name", "value")) |>
    dplyr::mutate(value = purrr::map_dbl(children, ~ sum(.x$value))) |>
    dplyr::rename(name = Industry) |>
    dplyr::left_join(colors_df, by = "name")

  min_valuation <- min(filtered_data$`Valuation ($B)`)
  max_valuation <- max(filtered_data$`Valuation ($B)`)

  treemap_data |>
    echarts4r::e_charts() |>
    echarts4r::e_title(
      text = paste0("Unicorns by Industry - from $",  min_valuation, "B to $", max_valuation, "B", "(", quartile,")")
    ) |>
    echarts4r::e_treemap(
      leafDepth = 1,
      itemStyle = list(
        normal = list(
          borderWidth = 0,
          gapWidth = 2,
          backgroundColor = "white"
        )
      ),
      upperLabel = list(
        normal = list(
          show = FALSE,
          height = 30,
          formatter = "{b}",
          color = "black",
          fontSize = 12
        )
      )
    ) |>
    echarts4r::e_tooltip(
      backgroundColor = "rgba(255,255,255,0.8)",
      formatter = treemap_tooltip
    )
}

Below we can see treemaps created in R, with the echarts4r library, with data split into quartiles.

How to navigate in those treemaps?

• Click on the desired Industry and see all the companies for that category.
• Check the grey bar below the treemap, and click in the first grey square to go back to the initial view of the treemap.

Code

htmltools::div(
  tree_map_chart(data = unicorn_data, quartile = "Q1", colors_df = colors_industry),
  htmltools::br(),
  tree_map_chart(data = unicorn_data, quartile = "Q2", colors_df = colors_industry),
  htmltools::br(),
  tree_map_chart(data = unicorn_data, quartile = "Q3", colors_df = colors_industry),
  htmltools::br(),
  tree_map_chart(data = unicorn_data, quartile = "Q4", colors_df = colors_industry),
)

Unicorn companies in Brazil (my country)

Code

unicorn_data |>
  dplyr::arrange(dplyr::desc(`Valuation ($B)`)) |>
  dplyr::filter(Country == "Brazil") |>
  dplyr::select(Country, name = Company, value = `Valuation ($B)`) |>
  tidyr::nest(children = c("name", "value")) |>
  dplyr::mutate(value = purrr::map_dbl(children, ~ sum(.x$value))) |>
  dplyr::rename(name = Country) |>
  dplyr::left_join(colors_country, by = "name") |>
  echarts4r::e_charts() |>
    echarts4r::e_title(text = "Unicorns in Brazil") |>
    echarts4r::e_treemap(
      leafDepth = 2,
      itemStyle = list(
        normal = list(
          borderWidth = 0,
          gapWidth = 2,
          backgroundColor = "white"
        )
      ),
      upperLabel = list(
        normal = list(
          show = TRUE,
          height = 30,
          formatter = "{b}",
          color = "black",
          fontSize = 12
        )
      )
    ) |>
    echarts4r::e_tooltip(
      backgroundColor = "rgba(255,255,255,0.8)",
      formatter = treemap_tooltip
    )

Final comments

This post showed a brief example of generating an interactive report based on Data Analysis. This post shows how we can facilitate the understanding of a data set by plotting charts in an objective way.

Below are some points I’d like to highlight about the technical side from this blog post:

• When doing data analysis, try to keep the same colors for categorical variables in the whole study. It is basic, but requires some extra work, like creating a palette color and setting the desired colors usually as a column in the dataset. But this is a very important step to allow the user to quickly reach some conclusions from the data.

• Treemaps are very nice, but sometimes they are not clear about the size of some categories. So, other charts, like the column charts we used here, can help in understanding better the behavior of some categories.

I hope you enjoyed this content. Please leave your comments below. If you want to see the updated version of this analysis, please also leave a comment.