Making legends specific to graphs within a figure using ggplot2 and gridExtra in R-CodePudding

I have the following plot:

df1 <- data.frame(col1 = rep(LETTERS[1:3], each = 4), col2 = rnorm(12), col3 = runif(12), col4 = rep(c("Fred", "Bob"), each = 6))
df1_list <- split(df1, df1$col1)
gridExtra::grid.arrange(grobs = mapply(function (x, y) {
  ggplot2::ggplot(x, ggplot2::aes(x = col2, y = col3, col = col4))  
    ggplot2::geom_point()  
    ggplot2::ggtitle(y)  
    ggplot2::scale_color_manual('Person', values = c('Fred' = 'red', 'Bob' = 'blue'))
}, x = df1_list, y = names(df1_list), SIMPLIFY = F))

On the first graph in this figure, all the points represent Fred, but in the legend, Bob still appears. Is it possible to include only the people who appear in each graph in the legend?

I want to make sure 'Fred' is always 'red' and 'Bob' is always 'blue', so using the scale_discrete_manual function's drop = T argument won't work.

Thanks!

CodePudding user response：

Here is a possible solution where the values parameter in scale_color_manual is dynamically created.

set.seed(1234)
df1 <- data.frame(col1 = rep(LETTERS[1:3], each = 4), col2 = rnorm(12), 
             col3 = runif(12), col4 = rep(c("Fred", "Bob"), each = 6))
df1$col4 <- factor(df1$col4)
df1_list <- split(df1, df1$col1)
cols <- setNames(c('red', 'blue'), c('Fred','Bob'))
gridExtra::grid.arrange(grobs = mapply(function (x, y) {
  x$col4 <- droplevels(x$col4)
  idx <- match(levels(x$col4), names(cols))
  ggplot2::ggplot(x, ggplot2::aes(x = col2, y = col3, col = col4))  
    ggplot2::geom_point()  
    ggplot2::ggtitle(y)  
    ggplot2::scale_color_manual('Person', values = cols[idx])
}, x = df1_list, y = names(df1_list), SIMPLIFY = F))

CodePudding user response：

Can you share your sessionInfo() so we can see the package versions you run? I run your code and this is what I get.