Reorder geom_bar from high to low when using stat=“count”

Question

I want to reorder a geom_bar graph from high to low when using stat="count" so that I can apply a fill.

I tried using geom_bar(aes(x = reorder(x, -stat(count)), fill = type) but it didn't work and threw the error "Error: stat_count requires the following missing aesthetics: x"

library(ggplot2)
df <- data.frame(x = c("Bob", "James", "Mary", "Sally", "Timmy", "Sally", "Sally", "Bob", "Bob", "Mary"), type = c("A", "B", "A", "B", "B", "C", "B", "B", "A", "B"))
ggplot(df) +
  geom_bar(aes(x = x, fill = type), stat = "count") +
  theme(axis.text.x = element_text(angle = 90, vjust = 0.5))

I want the bars to be order from highest count on the left to lowest count on the right.

Answer 1

I'm not sure about a ggplot2 solution, but I would tackle this using the forcats package. There is a function fct_infreq() to set the factor levels in order by frequency.

You could then do:

ggplot(df) +
    geom_bar(aes(x = forcats::fct_infreq(x), fill = type)) +
    theme(axis.text.x = element_text(angle = 90, vjust = 0.5))

Answer 2

Here is a solution with reorder from ggplot2 :

First, you need to count the occurence by names :

df2 <- df %>% group_by(x) %>%   mutate(count_name_occurr = n())

Then when specifying x axis, you reorder x by descending name occurences.

g2<-ggplot(data=df2, aes(x=reorder(x,-count_name_occurr), fill=type)) +
  geom_bar(stat="count")
g2