Reorder levels of factor by continuous months

Question

I have a data frame like so:

Dataframe:

 df<- data.frame(station= as.factor(rep("a", 12)), 
            month= as.character(c("Apr", "Aug", "Dec", "Feb", "Jan", "Jul",
                                                                  "Jun", "Mar", "May", "Nov", "Oct", "Sep")), 
            mean= rnorm(12, 300, 10), sd= rnorm(12, 300, 1), season= c(rep("wet", 6), rep("dry", 6)))

I want to reorder the months in consecutive order so that when I input into a ggplot function data for each month occurs in the calendar order.

Failed attempt to order:

df$month<-factor(df$month, levels=c("Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"))

I have previously got this to work and used the following code to get the subsequent output, however I erased something along the way and believe the source of error is with improperly ordering:

Previous outputs:

p<
  ggplot(df, aes(x = month, y=mean, fill=season)) + 
  geom_bar(stat="identity", color="black", 
       position=position_dodge()) +
  geom_errorbar(aes(ymin=mean-se, ymax=mean+se), width=.2,
             position=position_dodge(.9)) 
 p +labs(x="Month", y = "Precipitation (mm)") +
  theme_bw()+
  theme(legend.title = element_blank(),legend.text=element_text(size=15),legend.position="top",title= element_text(size=20), axis.text= element_text(size=10, face = "bold"), panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
   panel.background = element_blank(), axis.line = element_line(colour = "black")) +
  coord_flip()

I want to ultimately produce this figure again (note: for simplification, my example code only includes two seasons rather than the three in the image).

Is it that the problem that the order now goes from bottom to top instead of top to bottom? In that case, you could add `scale_x_discrete(limits = rev(levels(df$month)))` as shown [here](https://stackoverflow.com/questions/34227967/reversed-order-after-coord-flip-in-r). Otherwise your code works for me to put things in order by date (from bottom to top of plot, though). — aosmith, Nov 08 '18 at 19:16
FYI, you can use the built-in variable `month.abb` instead of typing out `"Jan", "Feb", ...`. — Gregor Thomas, Nov 08 '18 at 21:11

Jrakru56 · Answer 1 · 2018-11-08T21:07:30.853

0

You were almost there. Something like this should work :

df<- data.frame(station= as.factor(rep("a", 12)), 
                month= as.character(c("Apr", "Aug", "Dec", "Feb", "Jan", "Jul",
                                      "Jun", "Mar", "May", "Nov", "Oct", "Sep")), 
                mean= rnorm(12, 300, 10), se= rnorm(12, 10, 1), season= c(rep("wet", 6), rep("dry", 6)))



mon <- rev(c("Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sep", "Oct", "Nov", "Dec"))

p<- ggplot(df, aes(x = factor(month, levels= mon, labels = mon), y=mean, fill=season)) + 
  geom_bar(stat="identity", color="black", position=position_dodge()) +
  geom_errorbar(aes(ymin=mean-se, ymax=mean+se), width=.2, position=position_dodge(.9)) 

p +labs(x="Month", y = "Precipitation (mm)") +
  theme_bw()+
  theme(legend.title = element_blank(),legend.text=element_text(size=15),legend.position="top",title= element_text(size=20), axis.text= element_text(size=10, face = "bold"), panel.grid.major = element_blank(), panel.grid.minor = element_blank(),
        panel.background = element_blank(), axis.line = element_line(colour = "black")) + 
  coord_flip()

I am not sure why the behaviour is different if you only use level instead of level + label. Anyone?

edited Nov 08 '18 at 21:07

answered Nov 08 '18 at 19:03

Jrakru56

1,211
9
16

Thank you, this did order them properly. Only issue now is, if I choose to use `coord_flip() the top starts as Dec and bottom goes to Jan. Any idea how to order in such a way so that the topmost value is Jan an it sequences to Dec which would become the bottommost value? – Danielle Nov 08 '18 at 19:16
I am not sure of the proper way but a quick fix would be reverse the list i.e `mon2 <- rev(mon)` – Jrakru56 Nov 08 '18 at 19:21
Aha! The proper way is to use `scale_x_reverse()` – Jrakru56 Nov 08 '18 at 19:23
@Jrakru56 if `scale_x_reverse()` works update your answer to include that . – Mike Nov 08 '18 at 20:48
Hmm. It seems that `scale_x_reverse` is not the proper way. It does not handle discrete cases. So the quick fix way it is then ... – Jrakru56 Nov 08 '18 at 21:05

Reorder levels of factor by continuous months

1 Answers1