[Request]: Create a single row stacked column chart ordered by value with "other" as last entry

[Andrew-Saul]

Existing topic?

No - New topic.

What is the topic?

plots

What is the example

library(tidyverse)

# data
tib <- tibble(
  cat_name = c("alpha", "beta", "omega", "other", "epsilon", "zeta"),
  value = sample(1:50, 6, replace = TRUE)
) %>%
  add_count(wt = value, name = "total") %>%
  mutate(prop = value / total)

# basic plot
tib %>%
  ggplot(aes(fill = cat_name, y = prop, x = "")) +
  geom_col() +
  coord_flip()

#####################################################################
# how to change order of categories from increasing to decreasing value, "other" at end
# align legend order with plot
# write absolute values in plot

# order categories excluding "other" according to descending value
revised_levels <- tib %>%
  arrange(value) %>%
  filter(cat_name != "other") %>%
  pull(cat_name)

# add "other" at end of vector
revised_levels <- append("other", revised_levels)

# desired output
# desc order of factors by value with "other" last
# percentage value overlay factor
# legend at bottom, 1 row and in order seen on plot

tib %>%
  mutate(cat_name = factor(cat_name, levels = revised_levels)) %>%
  ggplot(aes(fill = cat_name, y = prop, x = "")) +
  geom_col() +
  # geom_bar(position = "stack", stat = "identity")+
  coord_flip() +
  scale_y_continuous(
    name = "percentage",
    labels = scales::label_percent(accuracy = 1)
  ) +
  geom_text(
    aes(label = scales::percent(prop)),
    position = position_stack(vjust = 0.5),
    size = 5
  ) +
  theme(legend.position = "bottom") +
  guides(fill = guide_legend(
    title = NULL,
    nrow = 1,
    reverse = TRUE
  ))

[Moohan]

Hey @Andrew-Saul thanks 😄 for reference in GitHub comments etc. You can use Markdown, so code blocks should start (and end) with three backticks ``` and adding R will do R syntax highlighting 💯 ```R

Can we make this a PR and develop it like that? I think the idea is solid, but in a PR, we can do an easier line-by-line code review!

Some thoughts:

• Load specific libraries, instead of tidyverse
• I think coord_flip has been superseded
• Use set.seed so the generated data will be the same
• Not sure but forcats may have functions to simplify messing about with factors - Alternatively could we do without, which would make the example simpler and more focused?

forcats::fct_reorder(cat_name, value, .desc = TRUE) %>% 
           forcats::fct_other(drop = "other") %>% 
           forcats::fct_rev()