Introduction

In the previous post, we start to plot maps with the collected data. However, we may only collect location data points or variables based on implicit geodata (e.g. countries). Therefore, we need external geodata like country boundaries as the base layers for our maps. In this blog post, we will dive into plotting maps with external geodata. We combine such external geodata with the dataset from an openwashdata package, and plot a thematic map.

Thematic map: what and why?

What is a thematic map? According to Wikipedia, we can think of thematic map as

A thematic map is a type of map that portrays the geographic pattern of a particular subject matter in a geographic area. This usually involves the use of map symbols to visualize selected properties of geographic features that are not naturally visible, such as temperature, language, or population. ¹

Useful R libraries:

rnaturalearth: An R package to hold and facilitate interaction with natural earth map data

Useful openwashdata datasets

gdho: Global database of humanitarian organizations from Humanitarian Outcomes.

Plot a thematic map with external geodata

In this demo, we will show you how to plot a map to reflect the headquarter concentration of the humanitarian headquarters around the world. At the end of the demo, you will learn how to:

Summarize headquarter information from gdho data
Combine gdho data with external geolocation data
Visualize a thematic map about headquarter location distribution

# Load libraries
library(gdho)
library(dplyr)
library(ggplot2)
library(sf) # for geolocation 
library(rnaturalearth) # for geolocation 
library(viridis)  
library(kableExtra) # for nice table display

1. Data processing: summarize headquarters

The data gdho contains the headquarter information of each humanitarian organization in the variable hq_location. First, we look at how many headquarters each country has. We process the column hq_location and display the top countries:

gdho_hq <- gdho |>
  filter(!is.na(hq_location)) |> # remove NA data 
  group_by(hq_location) |> # group by countries
  summarise(count = n()) |> # count the number of headquarters of each country
  arrange(desc(count)) # sort by descending order
  
gdho_hq |>
  head() |>
  kbl()

hq_location	count
United States	324
Somalia	266
Afghanistan	241
Pakistan	177
Colombia	169
Ukraine	165

We can see that the top six countries having the most humanitarian organization headquarters are United States, Somalia, Afgahnistan, Pakistan, Colombia and Ukraine.

2. Data combination: join with geodata

Now, we need to acquire geographical information of the countries (e.g. the borders). Here we use the package rnaturalearth and sf to help. The package rnaturalearth contains geodata about the countries in the world, and we import this dataset and combine it with our gdho data on the column hq_location.

world <- rnaturalearth::ne_countries(returnclass = "sf")

hq_map_data <- left_join(world, gdho_hq, by = c("name_long" = "hq_location"))

3. Map visualization: concentration of the headquarters

We have the data that contains all information we need: the counts of humanitarian organization headquarters each country has, and the geo-information of each country. We start by plotting the world view of the headquarter concentration.

hq_map <- ggplot() +
  theme_void() +
  geom_sf(data = hq_map_data, aes(fill = count), color = "white", lwd = 0)  # plot map
hq_map

Hooray! We got a map that reflects about the humanitarian headquarter concentration globally. The lighter blue indicates a larger number of headquarters, where we can see some countries like U.S. and Ukraine are on this side. That is matching our findings in Section 2.1.

But this map is somehow hart to read and lack of description to stand alone. We will polish the map by grouping the countries into several groups depending on their headquarter numbers and add more explanations.

3.1 Rescale the data

To better visualize the concentration, we rescale the data and set up groups for the number of headquarters, i.e. countries having 1-10 headquarters, having 10-50 headquarters and etc.. A useful trick is to transform the data via a log transformation so the color will be set up more distinguishable. We use scale_fill_viridis to achieve this as the following:

hq_map <- hq_map +
    scale_fill_viridis(trans = "log", breaks=c(1,10,50,100,300), 
                       # Legend style
                       name = "Num. Headquarters",
                       guide = guide_legend(
                           keyheight = unit (2, units = "mm"),
                           label.position = "bottom", 
                           title.position = 'top', nrow = 1))
hq_map

For comparison, if we do not apply the log transformation trans = "log", it will look the following:

hq_map +
    scale_fill_viridis(breaks=c(1,10,50,100,300))

3.2 Provide clear description

We add title and caption to explain the figure.

hq_map <- hq_map +
    labs(
    title = "Humanitarian Organization Headquarters Concentration",
    subtitle = "Number of headquarters per country"
  )  # add metadata
hq_map

3.3 Format the theme

Almost done! The final polish is to make it more visually appealing. For instance, we do not have headquarters on Antarctic, so removing it from the map is good to reduce unnecessary information. We further provide a grey background and reposition the legend.

hq_map <- hq_map +
  theme(
    plot.background = element_rect(fill = "#f5f5f2", color = NA),
    legend.position = c(0.13, 0.15)
  ) + # polish the appearance
  coord_sf(ylim = c(-50, 90)) # Remove antarctic
hq_map