Nested categorical variables as aesthetics-CodePudding

Say I have a dataset where I have a categorical variable where the categories are of two different fundamental types: for instance, that one of them represents the total of the values from the other two categories. I can encode this by a separate categorical variable, say is_total.

library(tidyverse)

dat <- tribble(
  ~x, ~y, ~categorical, ~is_total,
   0,  0,          "A",     FALSE,
   1,  1,          "A",     FALSE,
   0,  1,          "B",     FALSE,
   1,  3,          "B",     FALSE,
   0,  1,      "Total",      TRUE,
   1,  4,      "Total",      TRUE
)

I then want to create a line plot, where the total is shown with one type of line and the two components are shown with another type of line, and all three lines have different colours. I can do that like this:

ggplot(dat, aes(x, y, colour = categorical, linetype = is_total))  
  geom_line()

The problem is that I don't want two legends: I want "Total" to be shown with a dashed line in the second legend (and then I can get rid of the legend for is_total using guides(linetype = FALSE)). How would I do that? (I hoped changing the titles to be the same using labs would make it work things out, but sadly not.)

CodePudding user response：

One option would be to map categorical on linetype too and set your desired linotypes via scale_linetype_manual:

library(tidyverse)

dat <- tribble(
  ~x, ~y, ~categorical, ~is_total,
  0,  0,          "A",     FALSE,
  1,  1,          "A",     FALSE,
  0,  1,          "B",     FALSE,
  1,  3,          "B",     FALSE,
  0,  1,      "Total",      TRUE,
  1,  4,      "Total",      TRUE
)

ggplot(dat, aes(x, y, colour = categorical, linetype = categorical))  
  geom_line()  
  scale_linetype_manual(values = c(A = "solid", B = "solid", Total = "dashed"))