如何在R中的ggplot2的geom_errorbar中使用不同的过滤器

Question

Using this sample data frame:

> dput(coun2b)
structure(list(Camden = c(13.9933481152993, 17.5410199556541, 
26.0055432372506, 19.1064301552106, 9.05764966740577, 17.5321507760532
), Guilford = c(24.674715261959, 27.5097949886105, 25.4646924829157, 
22.2637813211845, 7.60227790432802, 17.9681093394077), years = 2012:2017, 
    Camden_ymin = c(12.4514939737261, 15.4927722105436, 22.5744436662436, 
    16.8415649174844, 7.45264839077184, 15.6645677387521), Guilford_ymin = c(23.2136204848819, 
    26.3627764588421, 23.8076842636931, 20.383805927254, 5.58799564906578, 
    16.2548749333076), Camden_ymax = c(15.5352022568726, 19.5892677007646, 
    29.4366428082575, 21.3712953929369, 10.6626509440397, 19.3997338133543
    ), Guilford_ymax = c(26.1358100390361, 28.6568135183788, 
    27.1217007021384, 24.143756715115, 9.61656015959026, 19.6813437455079
    )), class = "data.frame", row.names = c(NA, -6L))

which looks like this:

coun2b
    Camden  Guilford Camden_ymin Guilford_ymin Camden_ymax Guilford_ymax
1 13.99335 24.674715   12.451494     23.213620    15.53520      26.13581
2 17.54102 27.509795   15.492772     26.362776    19.58927      28.65681
3 26.00554 25.464692   22.574444     23.807684    29.43664      27.12170
4 19.10643 22.263781   16.841565     20.383806    21.37130      24.14376
5  9.05765  7.602278    7.452648      5.587996    10.66265       9.61656
6 17.53215 17.968109   15.664568     16.254875    19.39973      19.68134

and I use that dataframe with this:

library(tidyverse)

ggplot(coun2b, aes(x=years, Guilford, group=years)) + 
  labs(title = "Counts in Guilford, N.C.", 
       #caption="P. infestans range: 18 - 22 C; P. nicotianae range: 25 - 35 C; \"a\" Year with\nmost N.C. P. infestans reports (n=16); \"aa\" Year with most N.C. P. nicotianae reports (n=23)",
       y="Number of Days", x="Year" ) + geom_col( position = "dodge") +
  geom_errorbar(aes(ymin=Guilford_ymin, ymax=Guilford_ymax), position="dodge") + 
  theme(axis.text.x = element_text(face="bold"), axis.title.x = element_text(size=14), 
        axis.text.y = element_text(face="bold"), axis.title.y = element_text(size=14), 
        title = element_text(size=12)) +
  scale_x_continuous("Year", labels = plotscalex, breaks=plotscalex) +
  geom_hline(aes(yintercept = mean(Guilford[years %in% 2012:2016]),
                 linetype='Mean for 2012-2016')) +
  scale_linetype_manual(name="Legend", values=c("Mean for 2012-2016"=1) )

I create this barplot:

However, my complete dataset is actually larger and shaped differently, as long version. This is a sample of the long version:

> dput(samp1)
structure(list(years = c(2012L, 2012L, 2012L, 2013L, 2013L, 2013L, 
2014L, 2014L, 2014L, 2012L, 2012L, 2012L, 2013L, 2013L, 2013L, 
2014L, 2014L, 2014L), valu = c("mean", "ymin", "ymax", "mean", 
"ymin", "ymax", "mean", "ymin", "ymax", "mean", "ymin", "ymax", 
"mean", "ymin", "ymax", "mean", "ymin", "ymax"), name = c("Camden", 
"Camden", "Camden", "Camden", "Camden", "Camden", "Camden", "Camden", 
"Camden", "Guilford", "Guilford", "Guilford", "Guilford", "Guilford", 
"Guilford", "Guilford", "Guilford", "Guilford"), value = c(13.9933481152993, 
12.4514939737261, 15.5352022568726, 17.5410199556541, 15.4927722105436, 
19.5892677007646, 26.0055432372506, 22.5744436662436, 29.4366428082575, 
24.674715261959, 23.2136204848819, 26.1358100390361, 27.5097949886105, 
26.3627764588421, 28.6568135183788, 25.4646924829157, 23.8076842636931, 
27.1217007021384), county = structure(c(1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L), levels = c("Camden", 
"Guilford", "Pasquotank", "Wake"), class = "factor")), row.names = c(NA, 
-18L), class = c("tbl_df", "tbl", "data.frame"))

I tried using:

samp1 %>% filter(county == "Camden") %>% 
    ggplot( aes(x=years, y=value, group=years)) + 
    labs(title = "Number of Days in April-August with Suitable Weather for\nLate Blight Sporulation in Camden, N.C.", y="Number of Days", x="Year" ) + 
    geom_col(data=samp1 %>% filter(county=="Camden", valu=="mean"), aes(x=years, 
                                                                         y=value), position = "dodge") +
    geom_errorbar(data=samp1 %>% filter(county=="Camden"), 
                  aes(ymin=samp1 %>% filter(valu=="ymin"), ymax=samp1 %>% filter(valu=="ymax"), position="dodge")) + 
    theme(axis.text.x = element_text(face="bold"), axis.title.x = element_text(size=14), 
          axis.text.y = element_text(face="bold"), axis.title.y = element_text(size=14), 
          title = element_text(size=12)) +
    scale_x_continuous("Year", labels = plotscalex, breaks=plotscalex) +
    geom_hline(aes(yintercept = mean(Camden[years %in% 2012:2016]),
                   linetype='Mean for 2012-2016'))+
    scale_linetype_manual(name="Legend", values=c("Mean for 2012-2016"=1) )

as an attempt to create the same plot as above with the data in the long form. I get this error message:

Error in `geom_errorbar()`:
! Problem while computing aesthetics.
ℹ Error occurred in the 2nd layer.
Caused by error in `check_aesthetics()`:
! Aesthetics must be either length 1 or the same as the data (9)
✖ Fix the following mappings: `ymin` and `ymax`
Run `rlang::last_error()` to see where the error occurred.
Warning message:
In geom_errorbar(data = samp1 %>% filter(county == "Camden"), aes(ymin = samp1 %>%  :
  Ignoring unknown aesthetics: position

Because of the long form of this dataframe, I use filter 2x before I get to the geom_errorbar() . I don't think that's the problem, I just don't know how to filter correctly for ymin and ymax. I tried geom_errorbar(data=samp1 %>% filter(county=="Camden"), aes(ymin=samp1 %>% filter(county=="Camden", valu=="ymin"), ymax=samp1 %>% filter(county=="Camden",valu=="ymax"), position="dodge")) as well as what's in the code block above and I can't get it to work. How can I use the long form data, samp1 , to create a plot that is the same as the plot created when the data are wide? I'm using the long form because I will have to do a side-by-side barplot for multiple counties, while in this post, I'm just using one county.

Answer 1

You are making this much harder than it needs to be. What's wrong with a simple pivot to get your data into the correct format in the first place? The only wrangling you then need inside the plot code is to get the groupwise hline :

library(tidyverse)

sampl %>%
  pivot_wider(names_from = valu, values_from = value) %>%
  ggplot(aes(years, mean)) +
  geom_col() +
  geom_errorbar(aes(ymin = ymin, ymax = ymax), width = 0.25) +
  geom_hline(data = . %>% group_by(county) %>% summarize(mean = mean(mean)),
             aes(yintercept = mean), linetype = 2) +
  facet_grid(.~county) +
  theme_gray(base_size = 16) +
  theme(strip.background = element_blank(),
        strip.text = element_text(size = 20, face = 2))

Or, if you want to do one plot at a time:

sampl %>%
  pivot_wider(names_from = valu, values_from = value) %>%
  filter(county == "Guilford") %>%
  ggplot(aes(years, mean)) +
  geom_col() +
  geom_errorbar(aes(ymin = ymin, ymax = ymax), width = 0.25) +
  geom_hline(aes(yintercept = mean(mean)), linetype = 2) +
  theme_gray(base_size = 16) +
  ggtitle("Guilford")

Answer 2

You get the error because the second data frame is not in the appropriate format: by pivoting we could set ymin and ymax to columns: Then we could filter only once and apply the code:

library(tidyverse)
  samp1 %>% 
    pivot_wider(names_from = valu, 
                values_from = value) %>% 
    filter(county == "Camden") %>% 
    ggplot( aes(x=years, y=mean, group=years)) + 
    labs(title = "Number of Days in April-August with Suitable Weather for\nLate Blight Sporulation in Camden, N.C.", y="Number of Days", x="Year" ) + 
    geom_col(position = "dodge")+
    geom_errorbar(aes(ymin=ymin, ymax=ymax), position="dodge")+
    theme(axis.text.x = element_text(face="bold"), axis.title.x = element_text(size=14), 
          axis.text.y = element_text(face="bold"), axis.title.y = element_text(size=14), 
          title = element_text(size=12)) +
    geom_hline(aes(yintercept = mean(mean),
                   linetype='Mean for 2012-2016'))+
    scale_linetype_manual(name="Legend", values=c("Mean for 2012-2016"=1) )