How does ggplot2 evaluate factor levels? - r

Question

Consider this example dataframe:
library(tidyverse)
df <- tibble(item = c("Banana", "Ananas", "Apple", "Blueberry", "Orange",
"Spinach", "Cabbage", "Broccoli", "Carrot", "Eggplant"),
category = c(rep("Fruit", 5), rep("Vegetable", 5)),
n = c(57, 19, 14, 11, 8, 318, 70, 33, 31, 23))
First, I attribute factor levels to item by n. I also add a variable item_fct in order to display the factor levels of item in the output table for a better understanding of my problem:
df %>%
mutate(item = fct_reorder(item, n),
item_fct = as.numeric(item))
# A tibble: 10 x 4
item category n item_fct
<fct> <chr> <dbl> <dbl>
1 Banana Fruit 57 8
2 Ananas Fruit 19 4
3 Apple Fruit 14 3
4 Blueberry Fruit 11 2
5 Orange Fruit 8 1
6 Spinach Vegetable 318 10
7 Cabbage Vegetable 70 9
8 Broccoli Vegetable 33 7
9 Carrot Vegetable 31 6
10 Eggplant Vegetable 23 5
This output table and the factor levels make sense to me: Item is labeled from 1-10 by n in an ascending order.
But if I group it by category, I get the following output:
df %>%
group_by(category) %>%
mutate(item = fct_reorder(item, n),
item_fct = as.numeric(item))
# A tibble: 10 x 4
# Groups: category [2]
item category n item_fct
<fct> <chr> <dbl> <dbl>
1 Banana Fruit 57 5
2 Ananas Fruit 19 4
3 Apple Fruit 14 3
4 Blueberry Fruit 11 2
5 Orange Fruit 8 1
6 Spinach Vegetable 318 5
7 Cabbage Vegetable 70 4
8 Broccoli Vegetable 33 3
9 Carrot Vegetable 31 2
10 Eggplant Vegetable 23 1
Here, the factor levels are rather confusing to me. The items, depending on the group, share the same factor levels (1-5). I would still have expected levels from 1-10, but ordered by n depending on the respective group. Like this:
# A tibble: 10 x 4
item category n item_fct
<chr> <chr> <dbl> <dbl>
1 Banana Fruit 57 5
2 Ananas Fruit 19 4
3 Apple Fruit 14 3
4 Blueberry Fruit 11 2
5 Orange Fruit 8 1
6 Spinach Vegetable 318 10
7 Cabbage Vegetable 70 9
8 Broccoli Vegetable 33 8
9 Carrot Vegetable 31 7
10 Eggplant Vegetable 23 6
My Question:
Ultimately, it's hard for me to understand why ggplot2 produces the following plot with this code:
df %>%
group_by(category) %>%
mutate(item = fct_reorder(item, n),
item_fct = as.numeric(item)) %>%
ggplot(aes(n, item, fill = category)) +
geom_col()
If the items are factored from 1-5 in their respective groups and thus share the same overall factor levels, then why does ggplot2 display the bars by the grouped variable?
This plot is actually my desired outcome, but I don't understand the logic behind it. Considering the factor levels, I would expect bars of the same factor level to be next to each other (basically starting with "Eggplant", then "Orange", then "Carrot", then "Blueberry" etc.).
I would've assumed that the following table would produce the plot, because here the factor levels are ordered in the same way the bars are ordered in the plot:
# A tibble: 10 x 4
item category n item_fct
<chr> <chr> <dbl> <dbl>
1 Banana Fruit 57 5
2 Ananas Fruit 19 4
3 Apple Fruit 14 3
4 Blueberry Fruit 11 2
5 Orange Fruit 8 1
6 Spinach Vegetable 318 10
7 Cabbage Vegetable 70 9
8 Broccoli Vegetable 33 8
9 Carrot Vegetable 31 7
10 Eggplant Vegetable 23 6