This tutorial walks through visualising data using ggplot2 faceting and the geofacet package for improved spatial faceting.
geofacet?The geofacet package is an extension to ggplot2 that allows you to facet plots geographically, rather than alphabetically or by some pre-determined factor organisation. This means that subplots can be arranged in a layout that reflects their relative locations in a country or region—improving interpretability when exploring spatial patterns.
Geofaceting arranges a sequence of plots of data for different geographical entities into a grid that strives to preserve some of the original geographical orientation of the entities.
Use built-in geographic grids (https://hafen.github.io/geofacet/reference/index.html)
Customize or build your own grids for any administrative layout (helpful tool here we will discuss later: https://hafen.github.io/grid-designer/)
Integrates seamlessly with facet_geo() to replace facet_wrap()
First install the package and then load the libraries into your R session.
# install.packages("geofacet")
library(geofacet)
library(ggplot2)
library(dplyr)
Attaching package: 'dplyr'The following objects are masked from 'package:stats':
filter, lagThe following objects are masked from 'package:base':
intersect, setdiff, setequal, unionlibrary(scales)The example data for this tutorial can be found in the Box Folder: Data fellowship common folder > secondary-r-skills > data-viz > data
case_data <- readRDS("data/geofacet-data.rds")Lets take a look at the data first.
head(case_data)# A tibble: 6 × 6
state period total_cases year population confirmed_inc_per_1000
<chr> <date> <dbl> <dbl> <dbl> <dbl>
1 Abia 2020-01-01 3227 2020 3878977 0.832
2 Abia 2020-02-01 4118 2020 3878977 1.06
3 Abia 2020-03-01 3476 2020 3878977 0.896
4 Abia 2020-04-01 2987 2020 3878977 0.770
5 Abia 2020-05-01 3604 2020 3878977 0.929
6 Abia 2020-06-01 4726 2020 3878977 1.22 This dataset includes:
State-level malaria case counts (total_cases)
A period column indicating time by month
A year column
A population column
Incidence per 1,000 population (confirmed_inc_per_1000)
Let’s determine the time span covered:
min(case_data$year)[1] 2020max(case_data$year)[1] 2023Now lets plot these total case counts in a time series plotting all of our States in a single plot.
ggplot(case_data, aes(x = period, y = confirmed_inc_per_1000, color = state)) + # define plotting values from the data
geom_line() + # make a line chart
labs(y = "Incidence (per 1,000 population)", x = "Date",
title = "Total malaria case incidence by State 2020–2023") + # update the text labels
scale_color_viridis_d("state") + # use viridis colour palette
theme_minimal(base_size = 14) + # set theming
scale_y_continuous(labels = comma) + # use clean yaxis numeric labels
theme(axis.text.x = element_text(angle = 45, hjust = 1)) # rotate x axis for clarity 
What you will see here is that there is alot of information contained in this plot and it’s hard for us to distinguish within and between the different states. In addition the colour legend takes up a large proportion of the plot space making it harder to read.
Faceting allows us to break out this large plot into individual sub plots for each State with a single line of code facet_wrap(~ state, labeller = label_wrap_gen(width = 20, multi_line = TRUE)) and the can turn the legend off as each plot has its own State title.
ggplot(case_data, aes(x = period, y = confirmed_inc_per_1000, color = state)) +
geom_line(show.legend = FALSE) + # hide legend
facet_wrap(~ state, labeller = label_wrap_gen(width = 20, multi_line = TRUE)) + # facet individual plots
labs(y = "Incidence (per 1,000 population)", x = "Date",
title = "Total malaria case incidence by State 2020–2023") +
scale_color_viridis_d("state") +
theme_minimal(base_size = 14) +
scale_y_continuous(labels = comma) +
theme(axis.text.x = element_text(angle = 45, hjust = 1)) 
Now each state has its own subplot, which is easier to interpret. However, notice that these subplots are still ordered alphabetically and don’t reflect the geographic arrangement of states in Nigeria. We could plot a series of maps that contain this information to assess spatial trends but this would result in over 12 months * 4 years = 48 maps which is alot.
Instead we can use geofacet to arrange facets in a grid that mirrors their geographic location.This helps highlight spatio/temporal trends.
geofacet includes a built-in Nigerian state grid (ng_state_grid1). This arranges states in a rough geographic layout.
# Preview the Nigerian state grid layout
geofacet::ng_state_grid1 |>
ggplot() +
theme_void() +
facet_geo(~ name, grid = "ng_state_grid1")
This plot confirms the grid structure before using it in our malaria plot.
Sometimes, the naming conventions in the geofacet grid don’t match your dataset exactly. You’ll need to harmonize these.
#save grid to object
ng_state_grid <- geofacet::ng_state_grid1
# check for discordant names
anti_join(tibble::as_tibble(ng_state_grid), case_data, by= c("name" = "state"))# A tibble: 3 × 4
row col code name
<dbl> <dbl> <chr> <chr>
1 3 3 NG.FC Abuja FCT
2 3 4 NG.NA Nassarawa
3 7 5 NG.AK Akwa Ibom# adjust those discordant names to match case_data
ng_state_grid$name[ng_state_grid$name == "Akwa Ibom"] <- "Akwa-Ibom"
ng_state_grid$name[ng_state_grid$name == "Abuja FCT"] <- "Federal Capital Territory"
ng_state_grid$name[ng_state_grid$name == "Nassarawa"] <- "Nasarawa"Let’s use facet_geo() to visualize state-level malaria data with Nigeria’s spatial layout.
ggplot(case_data, aes(x = period, y = confirmed_inc_per_1000, color = state)) +
facet_geo(~ state, grid = ng_state_grid, label = "name",
labeller = label_wrap_gen(width = 20, multi_line = TRUE)) +
geom_line(show.legend = FALSE, linewidth = 1) +
labs(y = "Incidence (per 1,000 population)", x = "Date",
title = "Total malaria case incidence by State 2020–2023") +
scale_color_viridis_d("state") +
theme_bw() +
scale_y_continuous(labels = comma, limits = c(0, NA)) +
theme(axis.text.x = element_text(angle = 45, hjust = 1),
strip.background = element_blank())
Each state’s line plot is now displayed in a layout that approximates Nigeria’s geography. This makes it easier to explore spatial trends—e.g., higher incidence in the north.
To better explore local trends, enable independent y-axis scales:
ggplot(case_data, aes(x = period, y = confirmed_inc_per_1000, color = state)) +
facet_geo(~ state, grid = ng_state_grid, label = "name",
scales = "free_y") + # free the y-axis scale
geom_line(show.legend = FALSE, linewidth = 1) +
labs(y = "Incidence (per 1,000 population)", x = "Date",
title = "Total malaria case incidence by State 2020–2023") +
scale_color_viridis_d("state") +
theme_bw() +
scale_y_continuous(labels = comma) +
theme(axis.text.x = element_text(angle = 45, hjust = 1),
strip.background = element_blank())
This way we can also determin certain patterns in the data - for example malaria transmission is more highly seasonal in the North East of the country, we can also see that for certain states incidence rates drop to 0 unexpecidly e.g. Lagos, Ogun, Rivers, Cross River and others which highlights some potential data quality issues that aren’t spatially correlated to explore.
If your country or subnational boundaries aren’t supported, you can create a custom grid through the package.
Creating your own grid is as easy as specifying a data frame with columns row and col containing unique pairs of positive integers indicating grid locations and columns beginning with name and code.
To design a grid you can launch this helpful application starting by visiting this link or from R by calling:
grid_design()This will open up a web application with an empty grid and instructions on how to fill it out. Basically you just need to paste in csv content about the geographic entities (the row and col columns are not required at this point).
For example if I wanted to recreate a grid of the regions of Senegal I would take the following steps:
.csv file into the text box
Then a grid of squares with these column attributes will be populated and you can interactively drag the squares around to get the grid you want. You can also add a link to a reference map to help you as you arrange the tiles.
Once you are happy you will have something like the image below - and the R Code from the tool can be copied and stored as your grid object to use with facet_geo()
custom_grid <-
data.frame(
row = c(2, 2, 3, 3, 3, 3, 4, 4, 4, 5, 6, 6, 6, 6),
col = c(2, 3, 1, 3, 4, 2, 2, 4, 3, 5, 5, 4, 3, 2),
code = c("LG", "SL", "DK", "DB", "MT", "TH", "FK", "KA", "KL", "TC", "KE", "KD", "SE", "ZG"),
name = c("Louga", "Saint-Louis", "Dakar", "Diourbel", "Matam", "Thies", "Fatick", "Kaffrine", "Kaolack", "Tambacounda", "Kedougou", "Kolda", "Sedhiou", "Ziguinchor"),
stringsAsFactors = FALSE
)
# this can now be used in grid = custom_grid inside facet_geo().Here are a few hands-on exercises to reinforce what you’ve learned:
Explore Built-in Grids:
Use available_grids() from the geofacet package to find out what grids are available.
Try creating a geofaceted plot using one of these built-in grids for another country (e.g., us_state_grid1, ke_county_grid1).
Use Your Own Country Grid:
Check if a grid exists for your country of interest using available_grids().
If a grid is available, load it with get_grid("<grid_name>") and use it to facet a dataset you have access to.
Build and Submit Your Own Grid:
If your country or region is not supported:
Build a grid manually using a data frame with name, code, row, and col.
Validate it using validate_grid().
Submit it to the geofacet GitHub repository for inclusion.
Use your custom grid in a new facet_geo() plot.
Apply Geofaceting to your plots!! 😊🎉