Label nested factor levels in ggplot

Question

My data are included below via dput and look like this:

> head(dat)
     General      Covariate       Beta     SE Season
1  Intercept      Intercept -4.0673077 0.0499 Summer
2 Reggedness SlopeVar_30Log  0.1835911 0.0130 Summer
3  Elevation         Ele_30  0.6194081 0.0186 Summer
4      Slope      Slope_500  2.3549884 0.0527 Summer
5    SlopeSq    Slope_500Sq -0.6729443 0.0326 Summer
6     CanCov    CanCov_1000 -0.3879034 0.0234 Summer

I have 8 estimates for Summer and 9 for Winter.

With these data I can make the following figure. In this instance I have General as the x-axis and each value within General is plotted on a single line and dodged. However, I want to label the multiple values of General by the Covariate name. Each level of General has two levles of Covariate

ggplot(dat [dat $General != "Intercept",], aes(y = Beta, x = General, color = Season)) + 
  geom_point(position=position_dodge(0.6), size = 5)+
  geom_errorbar(aes(ymin=Beta-SE, ymax=Beta+SE),width = 0.6, size = 1.2 , position=position_dodge(0.6))

If I simply change the x-axis to Covariate (as below) then each level is on a separate line, except where the levels are identical as with Ele_30.

ggplot(dat [dat $General != "Intercept",], aes(y = Beta, x = Covariate, color = Season)) + 
  geom_point(position=position_dodge(0.6), size = 3)+
  geom_errorbar(aes(ymin=Beta-SE, ymax=Beta+SE),width = 0.6, size = 1.2 , position=position_dodge(0.6))+
  theme(axis.text.x=element_text(angle=30, hjust=1))

Is it possible to plot each level of General on the same vertical line (as in the first figure above), but label each point according to Covariate (like in the bottom figure)? Said differently, I want the points plotted as in plot one, and a label under each dodged point that corresponds to Covariate.

Thanks for any help.

library(ggplot2)
dat <- structure(list(General = structure(c(3L, 5L, 2L, 6L, 7L, 1L, 
4L, 8L, 3L, 5L, 2L, 6L, 7L, 1L, 4L, 8L, 9L), .Label = c("CanCov", 
"Elevation", "Intercept", "NDVI", "Reggedness", "Slope", "SlopeSq", 
"Solar", "Snow"), class = "factor"), Covariate = structure(c(1L, 
2L, 4L, 5L, 7L, 9L, 11L, 13L, 1L, 3L, 4L, 6L, 8L, 10L, 12L, 14L, 
15L), .Label = c("Intercept", "SlopeVar_30Log", "VMR_30Log", 
"Ele_30", "Slope_500", "Slope_100", "Slope_500Sq", "Slope_100Sq", 
"CanCov_1000", "CanCov_30", "NDVI_MeanTin_1000", "NDVI_MeanAmp_1000", 
"RadSumm_30", "AspectCos_30", "SnowSWE_1000"), class = "factor"), 
    Beta = c(-4.06730774504316, 0.183591094124286, 0.61940812155272, 
    2.35498841169247, -0.672944306628242, -0.387903359930857, 
    -0.815906730978726, -0.351201997266259, -5.06483998950688, 
    0.454499369654063, -0.337997523846132, 2.10010144558015, 
    0.251723125039171, -0.281294023525884, -1.02515594862923, 
    -0.46350259164822, -0.498134174074173), SE = c(0.0499, 0.013, 
    0.0186, 0.0527, 0.0326, 0.0234, 0.0152, 0.0107, 0.113, 0.0126, 
    0.0255, 0.0427, 0.0228, 0.0222, 0.0271, 0.0197, 0.033), Season = structure(c(1L, 
    1L, 1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 
    2L), .Label = c("Summer", "Winter"), class = "factor")), .Names = c("General", 
"Covariate", "Beta", "SE", "Season"), row.names = c(NA, 17L), class = "data.frame")

You can use `x = interaction(Covariate,General)` or `...+ facet_grid(~General,scales = "free_x")` — Roman, Feb 23 '16 at 12:37
Cool tip. Nonetheless, each point is plotted on a separate vertical line when there are multiple levels of `Covariate`. Is it possible to have each point on a single vertical line and then `dodge`-ed? — B. Davis, Feb 23 '16 at 12:43
The second part of [this answer](http://stackoverflow.com/questions/20060949/ggplot2-multiple-sub-groups-of-a-bar-chart/20073020#20073020) perhaps? Or maybe [this Q&A](http://stackoverflow.com/questions/18165863/ggplot2-labels-of-grouping-values-below-the-plot), which provides several alternative of multi-row labels. — Henrik, Feb 23 '16 at 12:47
Are there cases where your factors `season` and `covariate` have both two variables. Then there should be 4 points "doged" together, right? — Roman, Feb 23 '16 at 13:14
@Jimbou and Henrik, thanks for the helpful comments. Given the links that Henrick provided maybe this should be marked as a duplicate...? Or Jimbou, do you want to post your first comment as an answer? — B. Davis, Feb 23 '16 at 18:36

Label nested factor levels in ggplot

0 Answers0