Radar or Exploding Pie Charts in R (fmsb)

I am trying to show how world population has grown and will grow in five regions. I thought I could abuse the radarchart from the fmsb package to do this, making each region a category, and then each year would be a connected polygon of 5 points plotted on the rays. Obviously, the following attempt was a failure.

library(fmsb)

rnames <- c("rest", "africa", "china", "oecd", "seasia")
y1960 <- c( 759,  283,   660,  815,   517)
y1990 <- c(1405,  630,  1177, 1109,  1006)
y2020 <- c(2041, 1341,  1439, 1369,  1605)
y2060 <- c(2533, 2905,  1333, 1446,  1934)
y2100 <- c(2441, 4280,  1065, 1400,  1689)

pop <- data.frame( y1960, y1990, y2020, y2060, y2100 )
colnames(pop) <- rnames

areas <- c(rgb(1, 0, 0, 0.25),
           rgb(0, 1, 0, 0.25),
           rgb(0, 0, 1, 0.25),
           rgb(0, 1, 1, 0.25),
           rgb(1, 0, 1, 0.25))

radarchart(pop,
           cglty = 1,
           cglcol = "gray",
           plwd = 2,
           plty = 1,
           pfcol = areas)

not my proudest moment, of course. I don't even know how radarchart is calculating points. easy fix?

Looking at this with a couple of packages: ggradar and fmsb. Seems to me the final appearance is related to how the data is normalised.

Alternative approach with ggradar

library(dplyr)
library(tidyr)
library(scales)
library(ggradar)

rnames <- c("rest", "africa", "china", "oecd", "seasia")
y1960 <- c( 759,  283,   660,  815,   517)
y1990 <- c(1405,  630,  1177, 1109,  1006)
y2020 <- c(2041, 1341,  1439, 1369,  1605)
y2060 <- c(2533, 2905,  1333, 1446,  1934)
y2100 <- c(2441, 4280,  1065, 1400,  1689)

# bit of wrangling to get the data into required shape...
pop_4gg <- 
  data.frame(rnames, y1960, y1990, y2020, y2060, y2100 ) |> 
  pivot_longer(starts_with("y"), names_to = "year", values_to = "pop") |> 
  pivot_wider(names_from = rnames, values_from = pop)|> 
  mutate(across(rest:seasia, rescale))

ggradar(pop_4gg)

^{Created on 2023-05-23 with reprex v2.0.2}

Added following OP's comment and revisiting a previous version using fmsb. Have tried two versions with differing approaches to setting maximum and minimum values.

library(fmsb)

areas <- c(rgb(1, 0, 0, 0.25),
           rgb(0, 1, 0, 0.25),
           rgb(0, 0, 1, 0.25),
           rgb(0, 1, 1, 0.25),
           rgb(1, 0, 1, 0.25))


# relative to population for each country

pop_by_country <- 
  rbind(summarise(pop, across(everything(), max)),
        summarise(pop, across(everything(), min)),
        pop)
  


radarchart(pop_by_country,
           cglty = 1,
           cglcol = "gray",
           plwd = 2,
           plty = 1,
           pfcol = areas)

# relative to max population
pop_by_all <- 
  rbind(rep(max(pop), 5),
        rep(min(pop), 5),
        pop)

radarchart(pop_by_all,
           cglty = 1,
           cglcol = "gray",
           plwd = 2,
           plty = 1,
           pfcol = areas)

^{Created on 2023-05-24 with reprex v2.0.2}