Is ```summarize()``` contractive with ```ggplot()```

Question

Hello everyone Here is one example code

library(palmerpenguins)
mypenguins = penguins %>% drop_na()
penguins_df <- mypenguins %>% 
  group_by(species) %>% 
  summarize(mean_g = mean(bill_length_mm)) 
penguins_df %>% 
  ggplot(aes(x = mean_g, y = species)) +
  geom_col(width = 0.5) +
  theme(
    axis.text.y = element_text(
      color = if_else(penguins_df$mean_g > 48, "red", "black"), 
      face = if_else(penguins_df$mean_g  > 48, "bold", "plain"), 
      size = 20
    )
  )

However when I want to integrate two parts "summarize" and "ggplot" like this

penguins_df <- mypenguins %>% 
  group_by(species) %>% 
  summarize(mean_g = mean(bill_length_mm)) %>%
  ggplot(aes(x = mean_g, y = species)) +
  geom_col(width = 0.5) +
  theme(
    axis.text.y = element_text(
      color = if_else(mean_g > 48, "red", "black"), 
      face = if_else(mean_g  > 48, "bold", "plain"), 
      size = 20
    )
  )

I was warned: "mean_g" not found, but I do define "mean_g" before, which confused me a lot. Would someone mind offering some advice kindly? Thanks.

`mean_g` is a column in your data.frame. When doing `if_else(mean_g > 48, ...)` R is looking for a variable called `mean_g` in the global environment not in the data you passed to ggplot2. TBMK there is no way to access the data passed to ggplot from within `theme()`. Your first code works as you tell ggplot2 to look for `mean_g` inside `penguins_df`. — stefan, Apr 14 '22 at 07:26
Also, vectorized input to `element_text()` is not officially supported, therefore you might need to use another package to color axis text selectively (such as the package `ggtext`) — benson23, Apr 14 '22 at 08:04

score 2 · Answer 1 · answered Apr 15 '22 at 02:45

We can use the "trick" of surrounding the ggplot expression with curly braces. The only downside is that we'll need to explicitly refer to the piped object with a dot.

Here's the code:

library(tidyverse)
library(palmerpenguins)


mypenguins <- penguins %>% drop_na()

penguins_df <- mypenguins %>%
  group_by(species) %>%
  summarize(mean_g = mean(bill_length_mm))

penguins_df %>%
  {
    ggplot(., aes(x = mean_g, y = species)) +
      geom_col(width = 0.5) +
      theme(
        axis.text.y = element_text(
          color = if_else(.$mean_g > 48, "red", "black"),
          face = if_else(.$mean_g > 48, "bold", "plain"),
          size = 20
        )
      )
  }

score 1 · Answer 2 · answered Apr 14 '22 at 09:17

Too long for a comment.

The aesthetics defined with ggplot(aes(...)), apply to geoms only, not to the theme. That's why this doesn't work.

So for instance, you can write:

mypenguins <- penguins %>% drop_na()
mypenguins %>% 
  group_by(species) %>% 
  summarize(mean_g = mean(bill_length_mm)) %>%
  ggplot(aes(x = mean_g, y = species)) +
  geom_col(width = 0.5)

This will work because mean_g is an aesthetic, so ggplot looks for it in the context of the data.frame (it looks for the mean_g column). The theme(...) function does not respect aesthetics, so it looks for mean_g in the calling environment, and fails because there is not such object.

Thank you guys so much! So the problem happened in tha fact that "mean_g" is not in the global environment, which is required for recognization of theme(). Right? — KaiLi, Apr 15 '22 at 01:24

Is ```summarize()``` contractive with ```ggplot()```

2 Answers2