How can I bin the legend of a continuous variable in ggplot2?

Question

I'm trying to make a choropleth map of counties in the US. My values have a large range, and are highly 0 inflated, so I would like to create custom bins for the legend and coloring. Ideally this would be using ggplot. I would also like to keep a standard structure for the legend so that I can use the same legend for multiple choropleths with different data.

The legend might be something 0, 1-20, 21-50, 51-150, 151-300, 301-800, with each "bin" having its color derived from Spectral or some other continuous color palette.

scale_color_steps() , scale_fill_brewer() and other similar functions, even with breaks and limits or n.breaks , don't seem to work. Any ideas?

library(choroplethr)
library(choroplethrMaps)
library(ggplot2)
data(county.regions)

df <- tibble(region = county.regions$region, value = c(rep(0,2000), rep(1,1000), sample(2:800, 143)))  

county_choropleth(df) +
  scale_colour_steps(palette='Spectral', n.breaks = 8)

Answer 1

The simple solution to achive your desired result is to set num_colors = 1 in country_choropleth .

From the docs of country_choropleth :

num_colors The number of colors to use on the map. A value of 0 uses a divergent scale (useful for visualizing negative and positive numbers), A value of 1 uses a continuous scale (useful for visualizing outliers), and a value in [2, 9] will use that many quantiles.

By default num_colors = 7 which means that the continuous data gets discretized so that adding a continuous fill scale throws an Error: Binned scales only support continuous data. Setting num_colors = 1 the data don't gets binned and we can apply a custom continuous fill scale:

library(choroplethr)
library(choroplethrMaps)
library(ggplot2)
data(county.regions)

set.seed(42)

df <- tibble::tibble(region = county.regions$region, value = runif(3143, 0, 800))  

breaks <- c(0, 20, 50, 150, 300, 800)

county_choropleth(df, num_colors = 1) +
  scale_fill_stepsn(colors = RColorBrewer::brewer.pal(5, "Spectral"), breaks = breaks)

county_choropleth(df, num_colors = 1) +
  scale_fill_viridis_b(breaks = breaks)