R: Is there a way to add unused data levels in a ggplot legend?

Question

I am trying to present a map using ggplot on a factor-type vector going from 0 to 4 that I color coded. The vector is in the dataframe I named spat.dataframe and the vector is qt.

# plot
ggplot(spat.dataframe, aes(long,lat,group=group)) + # the data
  geom_polygon(aes(fill=as.factor(qt))) + # make polygons
  scale_fill_manual(values = c("0"="#F0F0F0","1"="green","2"="red","3"="blue","4"="purple"),
                    labels = c(paste0("white (",length(spat.dataframe[spat.dataframe$qt==0,]),")"),
                                 paste0("green (",length(spat.dataframe[spat.dataframe$qt==1,]),")"),
                                 paste0("red (",length(spat.dataframe[spat.dataframe$qt==2,]),")"),
                                 paste0("blue (",length(spat.dataframe[spat.dataframe$qt==3,]),")"),
                                 paste0("pink (",length(spat.dataframe[spat.dataframe$qt==4,]),")")),
                      drop=F,
                      name=NULL) +
  theme(line = element_blank(),  # remove the background, tickmarks, etc
          axis.text = element_blank(),
          axis.title = element_blank(),
          panel.background = element_blank()) +
  ggtitle(title) + 
  geom_path( colour = "#6b6b6b", size = .5 ) +
  coord_equal()

My problem lies with the legend portion, because what I want to do is display all of the legend options even if there arent any listed in the vector. So the values with colors blue and purple as well. It may seem like a bit of a stretch to try but I want it on my figure so that I can also display the number of values possible and the number of polygons associated with that value. So, for the values 3 and 4 in my vector qt, both are zero (i.e., blue (0) and purple (0)).

Any suggestions would be greatly appreciated. Thanks.

If the values exist but you want to hide them, you can set the color for those values to a value with an alpha (two extra digits on a hexadecimal color) of `00`, i.e. completely transparent, e.g. `ggplot(mtcars, aes(wt, mpg, color = factor(cyl))) + geom_point() + scale_color_manual(values = c('blue', '#00000000', 'red'))`. If the values don't exist, I suppose you could hack a couple fake ones in to make the levels show up, but there may be a better way. — alistaire, Oct 28 '16 at 22:10
Thanks for the response! My basic goal was to display the options regardless of whether it was on the map. On another note I expect I will eventually have to use your first suggestion on a different problem later down the road so thanks for that. You were right that I would have to find some way to hack it in, and I think Zach's answer did just that. Thanks again! — JellisHeRo, Oct 28 '16 at 22:31

Zach · Accepted Answer · 2016-10-28T22:37:23.373

Here's a working example with mtcars, taken from the ggplot2 documentation here. Namely, you can use scale_color_manual with the limits argument instead.

Edit after comment below: the limits argument is what is important, you can pass it to both scale_color_manual or scale_fill_manual.

cols <- c("8" = "red","4" = "blue","6" = "darkgreen", "10" = "orange")
plot <- ggplot(mtcars, aes(x = mpg, y = wt, colour = factor(cyl))) + geom_point()
plot + scale_color_manual(values = cols,
                          limits = c("4", "6", "8", "10"),
                          labels = c("this", "is", "my", "test"))

That should be enough for you to adapt it to your problem. It's hard for me be sure this will work without having your actual data set, but here is a solution that should work:

cols= c("0"="#F0F0F0","1"="green","2"="red","3"="blue","4"="purple")

ggplot(spat.dataframe, aes(long,lat,group=group)) + # the data
  geom_polygon(aes(fill=as.factor(qt))) + # make polygons
  scale_fill_manual(limits = c("0", "1", "2", "3", "4")
                    values = cols,
                    labels = c(paste0("white (",length(spat.dataframe[spat.dataframe$qt==0,]),")"),
                               paste0("green (",length(spat.dataframe[spat.dataframe$qt==1,]),")"),
                               paste0("red (",length(spat.dataframe[spat.dataframe$qt==2,]),")"),
                               paste0("blue (",length(spat.dataframe[spat.dataframe$qt==3,]),")"),
                               paste0("pink (",length(spat.dataframe[spat.dataframe$qt==4,]),")")),
                     drop=F,
                     name=NULL) +
  theme(line = element_blank(),  # remove the background, tickmarks, etc
    axis.text = element_blank(),
    axis.title = element_blank(),
    panel.background = element_blank()) +
  ggtitle(title) + 
  geom_path( colour = "#6b6b6b", size = .5 ) +
  coord_equal()

Thanks for the response! Your example worked, the only difference is since I filled the polygons in by the values used, I have to use `scale_fill_manual` rather than `scale_color_manual`, which would give me a different colored in graph. Just for future reference if anyone else wants to know. Thanks again! — JellisHeRo, Oct 28 '16 at 22:34

R: Is there a way to add unused data levels in a ggplot legend?

1 Answers1