Wrong axis labels for ggplot facet grid

问题:

I was answering this question that I faced a strange problem;

I want to make a bar plot with ggplot2 and have x axis labels as the last character of Group column. Use substring(Group, 3, 3) for this purpose:

substring(df$Group, 3, 3)
# [1] "1" "2" "3" "1" "2" "1" "2" "3" "4"

But when I use it in ggplot like below, it prints 1 instead of 4 at the last tick;

ggplot(data=df, aes(x=Group, y=Value)) +
  geom_bar(stat="identity") + 
  scale_x_discrete(labels = substring(Group, 3, 3), expand=c(0.1,0.1)) +
  facet_grid(~ substring(Group, 1, 1), space="free_x", scales="free_x", switch="x") + 
  theme_bw() +
  theme(strip.placement = "outside",
        strip.background = element_rect(fill=NA,colour="grey50"),
        panel.spacing=unit(0,"cm"))

I can get it to work by using labels =unique(substring(Group, 3, 3) but would someone explain what's going on?

Data:

  df <- structure(list(Group = structure(1:9, .Label = c("1_1", "1_2",  
  "1_3", "2_1", "2_2", "3_1", "3_2", "3_3", "3_4"), class = "factor"),  
      Value = c(-1.23, 2.34, 0.56, 1.87, -2.4, 5.54, -0.98, -2.31,  
      6)), .Names = c("Group", "Value"), row.names = c(NA, -9L), class = "data.frame")

# > df

#   Group Value 
# 1   1_1 -1.23 
# 2   1_2  2.34 
# 3   1_3  0.56 
# 4   2_1  1.87 
# 5   2_2 -2.40 
# 6   3_1  5.54 
# 7   3_2 -0.98 
# 8   3_3 -2.31 
# 9   3_4  6.00

回答1:

I think that @Gregor's answer is the way to go, but I can explain the error. Even if you add df$ before group, you get the behavior you described. You can see why if you don't let the scales vary:

ggplot(data=df, aes(x=Group, y=Value)) +
  geom_bar(stat="identity") + 
  scale_x_discrete(labels = substring(df$Group, 3, 3), expand=c(0.1,0.1)) +
  facet_grid(~ substring(Group, 1, 1), switch="x") + 
  theme_bw() +
  theme(strip.placement = "outside",
        strip.background = element_rect(fill=NA,colour="grey50"),
        panel.spacing=unit(0,"cm"))

gives:

As you can see, here it gives the "4" correctly. However, when you are setting the labels directly, it is setting them (in order) for each facet separately when you allow the scales to vary. In essence, you are setting the labels to "1, 2, 3, 1, 2, 1, 2, 3, 4" each time, but only using the first few.

If you want to stick closer to your current, you'll also need to set the breaks, e.g.:

ggplot(data=df, aes(x=Group, y=Value)) +
  geom_bar(stat="identity") + 
  scale_x_discrete(breaks = df$Group, labels = substring(df$Group, 3, 3), expand=c(0.1,0.1)) +
  facet_grid(~ substring(Group, 1, 1), space="free_x", scales="free_x", switch="x") + 
  theme_bw() +
  theme(strip.placement = "outside",
        strip.background = element_rect(fill=NA,colour="grey50"),
        panel.spacing=unit(0,"cm"))

Which gives

回答2:

Let's simplify and start in a fresh R session:

df <- structure(list(Group = structure(1:9, .Label = c("1_1", "1_2",  
                                                       "1_3", "2_1", "2_2", "3_1", "3_2", "3_3", "3_4"), class = "factor"),  
                     Value = c(-1.23, 2.34, 0.56, 1.87, -2.4, 5.54, -0.98, -2.31,  
                              6)), .Names = c("Group", "Value"), row.names = c(NA, -9L), class = "data.frame")

library(ggplot2)
ggplot(data=df, aes(x=Group, y=Value)) +
    geom_bar(stat="identity") + 
    scale_x_discrete(labels = substring(Group, 3, 3), expand=c(0.1,0.1))
# Error in substring(Group, 3, 3) : object 'Group' not found

The scale_x_discrete arguments don't have access to the within-data environment. (I think there are only two places you can safely reference data column in ggplot2: inside aes() and as part of a formula in a faceting call.) Your code must be finding a Group object from your global environment, which is probably the whole column. This is why you aren't getting an error in your question, but I am when I use a fresh R session.

Usually the easiest way to make ggplot2 do what you want is to fix your data so that it matches what you want:

df$group_last = substring(df$Group, 3, 3)
ggplot(data=df, aes(x=group_last, y=Value)) +
    geom_bar(stat="identity") + 
    scale_x_discrete(expand=c(0.1,0.1))

At this point you can add your facets and theming back in and everything will work.