GGplot scale_fill_discrete not creating a legend

Question

I used the code in a prior stackoverflow post, which previously made me the graph i wanted, with a legend, however now i am using the exact same code yet i am not getting a legend on my barplot.

dput(year.dat2)
structure(list(year = structure(c(1136044800, 1167577200, 1199113200, 
1230735600, 1262275200, 1136044800, 1167577200, 1199113200, 1230735600, 
1262275200, 1136044800, 1167577200, 1199113200, 1230735600, 1262275200
), class = c("POSIXct", "POSIXt"), tzone = ""), variable = structure(c(1L, 
1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 3L, 3L, 3L, 3L, 3L), .Label = c("SM1", 
"SM2", "SM3"), class = "factor"), value = c(1.24863586758821, 
1.23185757914453, 1.10997352162401, 1.13683917747257, 0.987520605867152, 
2.21498726809749, 1.6378992693761, 1.25635623380691, 1.13585705516765, 
1.10169569342842, 7.40955802109858, 5.7940698875978, 6.03438772314438, 
6.82271157830123, 7.24402375195127)), row.names = c(NA, -15L), .Names = 
c("year", 
"variable", "value"), class = "data.frame")

ggplot(year.dat2, aes(x = year, y = value, fill = factor(variable))) +
  geom_bar(stat = "identity", position = "dodge")+
  scale_fill_discrete(name = "variable",
                      breaks = c(1, 2, 3),
                      labels = c("SM1", "SM2", "SM3")) +
  xlab("year") + 
  ylab("yearly Sub Mean")

Resulting plot:

Answer 1

Remove breaks in scale_fill_discrete , they do not correspond to values of your factored variable data used as fill aes .

This is the code you want:

ggplot(year.dat2, aes(x = year, y = value, fill = factor(variable))) +
  geom_bar(stat = "identity", position = "dodge") +
  scale_fill_discrete(name = "variable",
                  labels = c("SM1", "SM2", "SM3")) +
  xlab("year") + 
  ylab("yearly Sub Mean")

Note 1: You do not even need the labels parameter as you are not renaming the variable categories. scale_fill_discrete(name = "variable") would be sufficient or labs(fill="variable") is all you need to change the legend title.

Note 2: In your original post you linked to this SO question: How to get a barplot with several variables side by side grouped by a factor

Here, the breaks in the sample code scale_fill_discrete(name="Gender", breaks=c(1, 2), labels=c("Male", "Female")) actually references the values of gender in the original df. Meanwhile, labels was used to rename 1 to Male and 2 to Female in the legend.

  Group.1      tea     coke     beer    water gender
1       1 87.70171 27.24834 24.27099 37.24007      1
2       2 24.73330 25.27344 25.64657 24.34669      2

On the other hand, there are no 1,2,3 values in your variable data that would match the breaks you have set in your original code, and that is why your legend is not plotted.

And for fun, here is an example how you might use breaks and labels in your data set:

ggplot(year.dat2, aes(x = year, y = value, fill = factor(variable))) +
  geom_bar(stat = "identity", position = "dodge")+
  scale_fill_discrete(name = "variable",
                  breaks = c("SM2", "SM3", "SM1"),
                  labels = c("SM2 new label", "SM3 new label", "SM1 new label")) +
  xlab("year") + 
  ylab("yearly Sub Mean")