Simple World Heatmaps in Julia

1st Step: Handling Our Data in Julia

The Packages

The following script lists all the packages in Julia and R for plotting heatmaps. Keep in mind that any code written in Julia between R""" and """ is interpreted as R code, which is an application of the package RCall. [note] RCall occasionally fails to recognize R's path after installing it. If this occurs, try specifying the path to R on your computer and then building the package. Each of these steps on my computer by running ENV["R_HOME"]="C:\\Program Files\\R\\R-4.3.0" and Pkg.build("RCall").

using DataFrames, RCall, CSV, Downloads

R"""
library(ggplot2)
library(svglite) #to save graphs in svg format, otherwise not necessary
"""

The (Fake) Data

5×2 DataFrame
 Row │ iso3    share_income
     │ String  Float64
─────┼──────────────────────
   1 │ ABW         0.694218
   2 │ AFG         0.48446
   3 │ AGO         0.450032
   4 │ AIA         0.169835
   5 │ ALB         0.487509

Now, we proceed to plot a world heatmap that represents the variable share_income.

3rd Step: Calling R to Plot the Map

Once we've computed our statistics in Julia and identified countries by their ISO codes, the final step is to use R to create the heatmap. To accomplish this, we must first define a file path in Julia, where our graphs will be stored. If you run the code I provided, the folder will be located at C:\maps\graph_folder.

isdir(joinpath("C:/", "maps")) || mkdir(joinpath("C:/", "maps")) #create 'maps' folder if it doesn't exist

graphs_folder = joinpath("C:/", "maps")

To use R from within Julia, we need to convert the DataFrame into R's format. This can be done in two ways. The first option is to directly send merged_df to R, which will create an R's DataFrame in R with the same name. However, this option is only recommended if you plan to further manipulate the data in R, which is not our case. The second option is to simply interpolate merged_df when we reference it in R.

Below, we provide the implementation of both options. After this, we'll exclusively use the second option, as it's simpler.

# we create the graph using RCall


R"""
#baseline code
our_gg <- ggplot() + geom_polygon(data = $(merged_df), 
                                  aes(x=long, y = lat, group = group, 
                                      fill=share_income)) 

#saving the graph
ggsave(filename = file.path($(graphs_folder),"file_graph0.svg"), plot = our_gg)
"""

@rput merged_df     # we send our merged dataframe to R

# we create the graph using RCall
R"""
#baseline code
our_gg <- ggplot() + geom_polygon(data = merged_df, 
                                  aes(x=long, y = lat, group = group, 
                                      fill=share_income)) 

#saving the graph
ggsave(filename = file.path($(graphs_folder),"file_graph0.svg"), plot = our_gg)
"""

We refer to this graph as the baseline case, since it uses the default layout provided by ggplot2. The difference between the approaches can be seen through the use of @rput or the interpolation of merged_df. In both options, though, graphs_folder is directly interpolated, indicating to R where to store the graphs (we could've alternatively used @rput to send the variable graphs_folder to R).

The fill option constitutes the only information you need to fill in if you use this code as a template. It indicates the variable to be plotted, which is share_income in our case. The rest of the code should remain the same, regardless of what you wish to plot. In particular, aes(x=long, y = lat, group = group) must be specified in the same way every time you use it, as it provides the necessary information for drawing a world map.

And that's it! You can use the code snippet as a template for your own graphs. The resulting graph is shown below.

As the baseline graph may not be visually appealing, we next provide further options to enhance its appearance. The final code we'll provide can also be used as a template that can be "copied and pasted", similar to the baseline script.

Final Code that Includes All Options

You can use the following code off the shelf to plot a world heatmap. First, execute the following lines in Julia, without making any changes.

R"""
user_theme <- function(){
  theme(
    panel.background = element_blank(),
    panel.border     = element_blank(),
    panel.grid.major = element_blank(),
    panel.grid.minor = element_blank(),

    axis.line    = element_blank(),
    axis.text.x  = element_blank(),
    axis.text.y  = element_blank(),
    axis.ticks   = element_blank(),
    axis.title.x = element_blank(),
    axis.title.y = element_blank()
      )
      } 
"""

Then, execute the script below by only making two modifications: insert the name of your Julia DataFrame at the option data, and indicate the variable to be plotted at fill. In our example, these options respectively correspond to merged_df2 (if we do not want to plot Antarctica, otherwise merged_df) and share_income.

merged_df2 = merged_df[.!(occursin.(r"Antarct", merged_df.short_name_country)),:]

R"""
#base code
our_gg <- ggplot() + geom_polygon(data = $(merged_df2), 
                                  aes(x=long, y = lat, group = group, 
                                      fill=share_income),
                                  color = "white", linewidth=0.05)
                                  
#we get rid of background, grid, and latitude/longitude
our_gg <- our_gg + user_theme()

#changing aspect ratio
our_gg <- our_gg + coord_fixed(1.3) 

#to change color scale (see color codes here https://derekogle.com/NCGraphing/resources/colors)
our_gg <- our_gg + scale_fill_gradient(low="lightskyblue", high="steelblue4", name = "Income Share")

#additional options for the legend of color scale 
our_gg <- our_gg + theme(
                         legend.position = "bottom",
                         legend.key.height = unit(0.5, "cm"),
                         legend.key.width = unit(2, "cm"))

#saving the graph
height <- 5
ggsave(filename = file.path($(graphs_folder),"file_graph01.svg"), plot = our_gg, width = height * 3, height = height)
"""