Use argument place holder within ggplot

Question

I am trying to use an argument place holder . within a ggplot(). But it doesn't work for some reason I am not entirely sure of.

What I am doing is this (using the sample data from ggplot2/the tidyverse):

library(tidyverse)
library(magrittr)

corr_eqn <- function(x, y, digits = 2) {
  corr_coef <-
    round(cor(x, y, use = "pairwise.complete.obs"), digits = digits)
  paste("r = ", corr_coef)
}


economics %>%
  filter(date >= "1990-11-01") %>%
  ggplot(aes(pop, unemploy)) +  
  geom_point()+
  annotate(geom = "text", x=-Inf, y=Inf, hjust=0, vjust=1,
           label = economics[economics$date>="1990-11-01",] %$% corr_eqn(pop, unemploy))

However, I want to reduce the command behind label to label = . %$% corr_eqn(pop, unemploy). I.e. I do not want to call economics[economics$date>="1990-11-01",] again as I have already filtered for this:

economics %>%
  filter(date >= "1990-11-01") %>%
  ggplot(aes(pop, unemploy)) +  
  geom_point()+
  annotate(geom = "text", x=-Inf, y=Inf, hjust=0, vjust=1,
           label = . %$% corr_eqn(pop, unemploy))

However, it doesn't work with the argument place holder .. What should I do instead?

Plus, if it would be possible to nat having to list pop and unemploy as seperate arguments in the corr_eqn fn again, this would be also amazing.

sry, should have made this clear. `economics` is example data from the tidyverse/ggplot2 — BeSeLuFri, May 21 '20 at 09:45

Allan Cameron · Accepted Answer · 2020-05-21T11:48:47.720

3

The problem is that annotate is not within the pipe, so . has no meaning there. The + operator in ggplot does not have the same function as the %>% in magrittr; in your code the pipe effectively stops at the call to ggplot(). The + operator will allow the next function to add various elements to the plot, but it won't in general allow you to access the data that was fed to the ggplot() call in the way you would with the %>% operator.

On the other hand, if you use geom_text instead of annotate, these problems vanish because you are accessing the variables in your subsetted data directly:

economics %>%
  filter(date >= "1990-11-01") %>%
  ggplot(aes(pop, unemploy)) +  
  geom_point() + 
  geom_text(aes(x = min(pop), y = max(unemploy), label = corr_eqn(pop, unemploy)), 
            hjust = 0, vjust = 1, size = 6)

edited May 21 '20 at 11:48

answered May 21 '20 at 11:11

Allan Cameron

147,086
7
49
87

1

an issue with this is that the text is written for each row of the data, hence it is a bit blurry. If using geom_text compare the text to using a dataframe with one row: `textDat = with(economics[economics$date >= "1990-11-01", ], data.frame(x=min(pop),y=max(unemploy))); lbl = with(economics[economics$date >= "1990-11-01", ],corr_eqn(pop, unemploy))` then plot `plt + geom_text(data=textDat, aes(x,y, label=lbl), hjust = 0, vjust = 3, size = 6)` (where plt if your plot). This is why annotate is the better tool here. – user20650 May 21 '20 at 12:37
2

seems oyu can use `, check_overlap = TRUE` now -- https://stackoverflow.com/questions/11618392/ggplot-text-printed-by-geom-text-is-not-clear#comment80056059_11618392 – user20650 May 21 '20 at 12:46
2

@user20650 yes, you're right on both counts. The previous method I used to solve this problem was to add an `alpha = c(1, rep(0, length(var) - 1)` argument to the aesthetic mapping and `+ scale_alpha_identity()` to the plot, but `check_overlap` is a lot easier to use. – Allan Cameron May 21 '20 at 12:49

Use argument place holder within ggplot

1 Answers1

Linked