ggplot2::scale_fill_identity() is not applying fills as mapped in dataframe-CodePudding

I am plotting data with ggplot in R and trying to map colors to fruits through a column in the dataframe. I expect scale_fill_identity() to map these fills to geom_col(), but it switches the colors of two fruits. Reproducible snippet below:

f <- tibble(
  fruit=factor(c('banana','apple','clementine'),c('clementine','banana','apple')),
  value=c(20, 30, 50),
  color=c('yellow','red','orange'),
)

# table is as-expected
# bananas are yellow, apples are red, clementines are orange
f

# factors ordered properly
f |> count(fruit)

# auto colors look fine
f |> ggplot(aes('x', value, fill=fruit))   geom_col()

# bananas are red, apples are yellow
f |> ggplot(aes('x', value, fill=color))   geom_col()  
  scale_fill_identity(guide='legend', labels=levels(f$fruit))

# why are colors not matching the mapping i have in the table?
# should i be using something like breaks=f$color in the identity call?

What am I missing, shouldn't I be able to use map colors like this? I know I can provide a named vector in values of scale_fill_manual(), but I'd rather add values to the dataframe. Also, should I be using the breaks argument?

CodePudding user response：

You should add your breaks of your "color" column and remove levels from your labels like this:

library(tibble)
library(ggplot2)
f <- tibble(
  fruit=factor(c('banana','apple','clementine'),c('clementine','banana','apple')),
  value=c(20, 30, 50),
  color=c('yellow','red','orange'),
)

f |> ggplot(aes('x', value, fill=color))   geom_col()  
  scale_fill_identity(guide='legend', labels=f$fruit, breaks = f$color)

^{Created on 2022-09-17 with reprex v2.0.2}

Reverse order legend using rev:

f |> ggplot(aes('x', value, fill=color))   geom_col()  
  scale_fill_identity(guide='legend', labels=rev(f$fruit), breaks = rev(f$color))

^{Created on 2022-09-17 with reprex v2.0.2}