ggplot2: Fix colors to factor levels - r

Question

I'm working on a larger project for which I am creating several plots in ggplot2. The plots are concerned with plotting several different outcomes across several different discreet categories (think: countries, species, types). I would like to completely fix the mapping of discrete types to colors such that Type=A is always displayed in red, Type=B is always displayed in blue, and so on across all plots irrespective of what other factors are present. I know about scale_fill_manual() where I can provide color values manually and then work with drop = FALSE which helps in dealing with unused factor levels. However, I find this extremely cumbersome since every plot will need some manual work to deal with sorting the factors in the right way, sorting color values to match factor sorting, dropping unused levels, etc.
What I am looking for is a way where I can map once and globally factor levels to specific colors (A=green, B=blue, C=red, ...) and then just go about plotting whatever I please and ggplot picking the right colors.
Here is some code to illustrate the point.
# Full set with 4 categories
df1 <- data.frame(Value = c(40, 20, 10, 60),
Type = c("A", "B", "C", "D"))
ggplot(df1, aes(x = Type, y = Value, fill = Type)) + geom_bar(stat = "identity")
# Colors change complete because only 3 factor levels are present
df2 <- data.frame(Value = c(40, 20, 60),
Type = c("A", "B", "D"))
ggplot(df2, aes(x = Type, y = Value, fill = Type)) + geom_bar(stat = "identity")
# Colors change because factor is sorted differently
df3 <- data.frame(Value = c(40, 20, 10, 60),
Type = c("A", "B", "C", "D"))
df3$Type <- factor(df3$Type, levels = c("D", "C", "B", "A"), ordered = TRUE)
ggplot(df3, aes(x = Type, y = Value, fill = Type)) + geom_bar(stat = "identity")

Answer 1

You could define your own custom scale, if you like. If you look at the source for scale_fill_manual,
scale_fill_manual
#> function (..., values)
#> {
#> manual_scale("fill", values, ...)
#> }
#> <environment: namespace:ggplot2>
it's actually quite simple:
library(ggplot2)
scale_fill_chris <- function(...){
ggplot2:::manual_scale(
'fill',
values = setNames(c('green', 'blue', 'red', 'orange'), LETTERS[1:4]),
...
)
}
df1 <- data.frame(Value = c(40, 20, 10, 60),
Type = c("A", "B", "C", "D"))
ggplot(df1, aes(x = Type, y = Value, fill = Type)) +
geom_col() +
scale_fill_chris()
df2 <- data.frame(Value = c(40, 20, 60),
Type = c("A", "B", "D"))
ggplot(df2, aes(x = Type, y = Value, fill = Type)) +
geom_col() +
scale_fill_chris()
df3 <- data.frame(Value = c(40, 20, 10, 60),
Type = c("A", "B", "C", "D"))
df3$Type <- factor(df3$Type, levels = c("D", "C", "B", "A"), ordered = TRUE)
ggplot(df3, aes(x = Type, y = Value, fill = Type)) +
geom_col() +
scale_fill_chris()

Answer 2

You could make a custom plot function (including scale_fill_manual and reasonable default colours) in order to avoid repeating code:
library(ggplot2)
custom_plot <- function(.data,
colours = c("A" = "green", "B" = "blue", "C" = "red", "D" = "grey")) {
ggplot(.data, aes(x=Type, y=Value, fill= Type)) + geom_bar(stat="identity") +
scale_fill_manual(values = colours)
}
df1 <- data.frame(Value=c(40, 20, 10, 60), Type=c("A", "B", "C", "D"))
df2 <- data.frame(Value=c(40, 20, 60), Type=c("A", "B", "D"))
df3 <- data.frame(Value=c(40, 20, 10, 60), Type=c("A", "B", "C", "D"))
df3$Type <- factor(df3$Type, levels=c("D", "C", "B", "A"), ordered=TRUE)
custom_plot(df1)
custom_plot(df2)
custom_plot(df3)

Answer 3

Another options is to make drop = F the default by defining the default colour scales as follows:
scale_colour_discrete <- function(...)
scale_colour_manual(..., drop = F)
scale_fill_discrete <- function(...)
scale_fill_manual(..., drop = F)
That way colours are always consistent for different factors.

Answer 4

make sure you convert that column into Factor first and then create a variable to store the color value for each factor...
df$color <- as.factor(df$color, levels = c(1, 0))
cbPallete <- c("1"= "green", "0"="red")
ggplot(data = df) + geom_bar(x = df$x,
y = df$y,
fill = df$color) +
scale_fill_manual(values = cbPallete)