Creating Hexbin Maps in R

Objective

The objective of this tutorial is to create a thematic hexbin map of U.S. higher education in RStudio. Users will learn how to create a hexbin map from a geospatial object and plot thematic data. In addition, this tutorial will demonstrate how to add and customize various cartographic elements to your map including symbolization, labelling, and map elements (e.g., title, legend). Users will have the opportunity to become more familiar with the R programming language as well as explore the spatial and cartographic capabilites of the software. This tutorial uses open-source software and data and will discuss the advantages and limitations of each. Finally, this tutorial contributes to the collection of Open Source GIS tutorials created by students at Carleton University.

Software Requirements: R, RStudio, spreadsheet software (e.g., Microsoft Excel)

Skills: Cartographic design, programming

Note: This tutorial assumes basic knowledge of the R programming language. This version of the tutorial was created using a Windows platform with R version 3.6.2.

Why Hexagons?

Thematic mapping generally visualizes statistical data by using irregularly shaped polygons. These shapes are often politically-defined (e.g., states, provinces, counties) and can vary widely in size and shape. This can lead to disparities in viewers' perceptions of a map and the spatial patterns it reveals. Alternatively, we can use regularly shaped features in order to normalize geographies for the purpose of thematic mapping. Equilateral trianges, squares, and hexagons are the three polygon shapes which can be used to create regularly shaped, evenly spaced grids. Hexagons are a good choice for visualization as the shapes 'nest' together and the edge effects of the grid shape reduce sampling bias. In addition, hexagons can be used to obscure sensitive source data (e.g., personal addresses).

Getting Started

Downloading the Software

The first step of this tutorial is downloading R and RStudio if they are not already installed on your device. R version 4.1.2 is the latest version of the software released in 2021. R is a widely used open-source software environment used for data manipulation and analysis (statistics, graphics, etc.). R is easily customizable and is executed line by line in a console.

For the purposes of this tutorial, we will be using an integrated development environment (IDE) called RStudio. This software provides users with a console, a syntax-highlighting editor, and a set of integrated tools for plotting and debugging R code. RStudio is available in open source (free) and commercial ($995/year) editions, and has both a desktop and server version. This tutorial only requires the open source edition.

The desktop version of RStudio can be downloaded HERE

Finding Data

When creating a hexagonal map, users have the option to create a hexbin map from (1) a geospatial object or (2) a list of coordinates. For the purposes of this tutorial, we will use an existing hexagon boundary file (.geojson) of the United States published by a member of the GIS community. The file includes a total of fifty-one hexagons (50 U.S. States and District of Columbia). Download the data in .geojson format and save to a new project folder. (Note: you are not required to create a CARTO account to access the data.)

The hexgrid file is available to download HERE.

Figure 2. Screenshot of the CARTO website - Select 'Download' (circled) to begin

The attribute data for this tutorial will be sourced from the United States (U.S.) Department of Education, National Center for Education Statistics. The data of interest is the state-level “Bachelor's Degrees in Science and Engineering Conferred per 1,000 Individuals 18-24 Years Old” from 2000 to 2019. This data is an important indicator of higher education attainment and bachelor's-level training. According to the U.S. Department of Education, Science and Engineering (S&E) fields include physical, life, earth, ocean, atmospheric, computer, and social sciences; mathematics; engineering; and psychology (excludes medical and technology fields). Download the data and save it to your project folder. Spend some time exploring the data and taking note of any observable trends.

The data is available to view and download HERE.

Figure 3. Screenshot of the National Science Board website - Select 'Downloads' (circled) to begin

Cleaning the Data

Before importing the attribute data into R, we will need to clean the data and save it as a text file. The data provided includes three tables: (1) the number of S&E Bachelor's Degrees conferred, (2) the population of cohort 18-24 years old, and (3) the number of degrees conferred per 1,000 individuals. For our choropleth map, we are most interested in the third table because the values have been normalized. Following the instructions below, we will prepare the data for import.

Open the downloaded data file (se-bachelors-degrees-per-1000-18-24-year-olds.xlsx) in a spreadsheet software (E.g., Microsoft Excel)
Select and delete the 'S&E bachelor's degrees' and 'Individuals 18-24 years old' tables
Select the 'Degrees/1,000 individuals 18–24 years old' table (columns AR-BK)
Click on the column border, hold down the shift key and drag across
Simplify the data by deleting all unnecessary rows (Note: this map does not include Puerto Rico)
Save the spreadsheet to your directory folder as a comma-seperated values (.csv) file

Navigating RStudio

The RStudio workspace consists of four panes:

(Top left): Script window (Create and work on multiple scripts; Write and run your code)
(Bottom left): R console (View the results of a command and error messages)
(Top right): Environment (View a list of your created variables); History
(Bottom right): Plots (View and save your graphs and maps); Packages (Explore available packages); Help

Figure 3: Screenshot of the RStudio Environment

At the start of your session:

Start a new R script by clicking on the icon in the top left corner of the script window or go to File > New > R script.
Save your script to your project folder by clicking the 'save' icon or go to File > Save As.
Set your working directory (the folder where R reads and saves files). File > Session > Set Working Directory > Choose Directory
Comment your code using the (#) symbol

setwd("~/FALL 2021/GEOM 4008/Data")

#4008 Tutorial 
#How to Create a Hexbin Map
#December 2021

TIP: RStudio allows users to change the theme of the IDE's user interface. The 'Modern' theme is the default, however you switch to a darker palette if you prefer. Learn how HERE.

Creating a Choropleth Map

In this section, we will follow a series of steps to install R packages, and import, prepare, and visualize our spatial and attribute data. Next, we will create a choropleth map that shows the number of science and engineering bachelor's degrees conferred in the United States in 2019. Feel free to explore the functions introduced in the below sections and experiment with different customizations (e.g., colour schemes).

Installing Packages

R enables users to install and load packages from the CRAN. A package is a shareable bundle of code, sample data, documentation and tests. R packages are an advantage of this piece of software as they allow users to easily share their code with others. In additional to the pre-installed base package, we will need to download additional packages for this tutorial. Click the link beside each package listed below to learn more. Install the required packages using the below script. Once installed, we have to call each of the pacakges from our library. Alternatively, you can use the 'Install Packages' dialog box which is accessible through the main menu (Tools > Install Packages).

tidyverse Learn more
geojsonio Learn more
RColorBrewer Learn more
sp Learn more
broom Learn more
rgeos Learn more

#Install required packages for this tutorial
install.package('tidyverse')
install.package('geojsonio')
install.package('RColorBrewer')
install.package('sp')
install.package('broom')
install.package('rgeos')


#Load the packages
library(tidyverse)
library(geojsonio)
library(RColorBrewer)
library(sp)
library(broom) 
library(rgeos)

TIP: If you want to learn more about a specific package or function, you can write a command to view the corresponding 'Help' page. In the script window you can run help() or type the name of the function preceded by a question mark in the R console, for example ?brewer.pal (see Figure 5). If the package or function is not found in your library, you can still access the documentation by using two question marks, for example ??brewer.pal. The R documentation will provide a description, a list of arguments, details, references, as well as a number of examples.

Figure 4. Screenshot of R Documentation

Importing Data

Before reading your data into R, ensure that your data are located in the folder that you set as your working directory.

To import the hexbin data, we will use the geojson_read() function. The data will now appear in the 'Global Environment' tab in the top-right pane. After the file has been imported, we will need to reformat our data using the mutate and gsub functions and fortify it using the tidy function in order to plot our map in the next steps.

#Import hexbins
hex <- geojson_read("us_states_hexgrid.geojson", what = "sp")

#Reformat the 'google_name' field
#This will remove the (United States) from each value
#E.g., Vermont (United States) will be changed to Vermont 
hex@data = hex@data %>% mutate(google_name = gsub(" \\(United States\\)", "", google_name))

#Fortify the data to create a data format output
#This format is needed to plot the map using the ggplot2 package
hex_fortify <- tidy(hex, region = "google_name") 

#Plot the hexbins
#I chose a green hue for the fill and white for the outline
ggplot () +
  geom_polygon(data = hex_fortify, aes( x = long, y = lat, group = group), fill="#a1dab4", color="#f7f7f7") +
  geom_text () +
  theme_void () +
  coord_map ()

Figure 5. Plot of hexagon boundaries

Next, we will import the education data using the read.csv function. Once imported, we can view the data table and delete any unnecessary rows that contain NA values.

#Import education data
#Note: my file name is bachelor.csv
bach <- read.csv("bachelor.csv", header=T)
#Remove rows and columns with NA values
bach <-bach[-c(53:54), ] #rows
bach <-bach[-c(22:63)] #columns
#View the data table
view(bach) 
#Verify the class of our data is numeric 
sapply(bach, class)

Figure 6. Table of attribute data

Merging Data Frames

In order to create our choropleth map, we need to join our spatial and attribute data together. Using the left_join function, we will perform the join by merging our data frames. The two fields that we will use to join the data are "id" and "state". Run the script and view the data table to verify that the join was successful.

#Perform spatial join 
hex_fortify <- hex_fortify %>% 
       left_join(. , bach, by=c("id"="State")) 
#Check if the join was successful
view(hex_fortify)

Symbolizing Data

Choropleth maps are a common type of thematic map that portray geographic patterns for areal units such as states and provinces. They generally consist of two to six colour symbols which represent a corresponding number of nonoverlapping classes for an intensity index (Monmonier, 2018).

To symbolize the attribute data in our choropleth map, we will use graduated colour symbology. By varying the colours of the hexagons, our map will show the quantiative difference in the number of Bachelor's Degrees awarded in Science and Engineering (per 1,000 indiviuduals) in the age cohort of 18-24 between the U.S. states in 2019. This visualization method will allow us to observe and analyse spatial patterns in the attribute data. We will classify the data into ranges and assign a specific colour to each of the classes. When selecting the size and total number of classes, it is important to explore the distribution and central tendencies descriptive of our data.

The ColorBrewer tool can be explored more HERE

#Explore descriptive statistics 
mean(bach$X2019)
[1] 25.27635
range(bach$X2019)
[1] 9.68 67.07

#Create bins
#These labels will appear on the legend
hex_fortify$bin <- cut( hex_fortify$X2019 , breaks=c(9, 20, 30, 50, 70), labels=c("9-20", "20-30", "30-50", "50+"))

#Select a color ramp 
#Display all colorblind-friendly ColorBrewer palettes
display.brewer.all(colorblindFriendly = TRUE)

#Choose a sequential ramp for our map 
#Where n = number of data classes 
my_palette <- brewer.pal(n=4, name="Greens")

Figure 4. Colourblind-friendly Brewer palettes.

Adding Map Elements

Labels can be added to our map to provide viewers with geographic reference information. This is especially important for our hexbin map where the U.S. states boundaries are not positioned or shown as they would appear on a political map. To add labels to the hexbin map, we must first calculate the centroid of each hexagon using the gCentroid function. Next, we will use the two-letter state abbreviations in the "id" field. The labels will be added to the plot using the geom_text function. The colour and size of the labels can also be changed.

#Add labels 
centers <- cbind.data.frame(data.frame(gCentroid(hex, byid=TRUE), id=hex@data$iso3166_2))

A legend can be created and added to our map to help the viewer understand what each colour symbol represents. The legend should include a title as well as labels that indicate the data range for each symbol. Different functions found within the ggplot2 package allow us to add a legend to our map and customize its size, type colour, background colour and position.

A title is another important cartographic element that is used to tell the viewer what the map is about. The title of our map should answer the questions of what?, where?, and when?.

#Plot our final map
ggplot() +
  geom_polygon(data=hex_fortify, aes(fill=bin, x=long, x=lat, group=group), size=0, alpha=0.9, color="#f7f7f7") +
  geom_text(data=centers, aes(x=x, y=y, label=id), color="#252525", size=5) + #this adds our labels
  theme_void() +
  scale_fill_manual(
    values=my_palette, 
    name="Degrees per 1,000 Individuals 18-24 Years Old",
    guide= guide_legend( keyheight=unit(4, units="mm"), keywidth=unit(10, units="mm"), direction="horizontal", label.position="bottom", title.position="top", nrow=1)
  ) +
  ggtitle( "Bachelor's Degrees in Science & Engineering Conferred in the United States in 2019" ) +
  theme(
    legend.position = c(0.5, 0.9),
    text = element_text(color = "#f7f7f7", face="bold"),
    plot.background = element_rect(fill = "#252525", color = NA),
    panel.background = element_rect(fill = "#252525", color = NA), 
    legend.background = element_rect(fill = "#252525", color = NA),
    plot.title = element_text(size=18, hjust=0.5, color = "#f7f7f7", face="bold"),
  )

Results

Figure 6. Final choropleth map

References

Esri. (2015, April 8). Thematic mapping with hexagons. https://www.esri.com/about/newsroom/insider/thematic-mapping-with-hexagons/

Holtz, Yan. (n.d.). Hexbin map in R: an example with US states. https://www.r-graph-gallery.com/328-hexbin-map-of-the-usa.html

https://team.carto.com/u/andrew/tables/andrew.us_states_hexgrid/public/map

Monmonier, M. (2018). How to lie with maps. (3rd ed.). The University of Chicago Press.

National Science Board. (2021). https://ncses.nsf.gov/indicators/states/indicator/se-bachelors-degrees-per-1000-18-24-year-olds

Resources

The R Graph Gallery: A collection of charts made with the R programming language [1]

ColorBrewer: A diagnostic tool for evaluating colour schemes for cartography [2]