堆叠条形图中每个 x 轴组的百分比标签

Question

我想使用

ggplot2

将每组的百分比标签添加到绘图中

带有总体（所有 x 轴组）百分比标签的示例

# library
library(reshape2)
library(tidyverse)

# make plot
p <- ggplot(tips, aes(x= day,  group=sex)) + 
  geom_bar(aes(y = ..prop.., fill = factor(..group..)), stat="count") +
  geom_text(aes( label = scales::percent(..prop..),
                 y= ..prop.. ), stat= "count", position=position_stack(vjust = 0.5)) +
  labs(y = "Percent", fill="sex") +
  scale_y_continuous(labels = scales::percent)

p

现在我想每天添加标签（即 x 轴组）：例如，1 为 52.6%，2 为 47.4%

Fri

我知道如何计算这些值，但是，我不知道如何在

ggplot

调用中执行此操作。

# make percentage labels per day
tips %>% 
  group_by(day,sex) %>% 
  summarise(total=n()) %>%
  left_join(tips %>% group_by(day) %>% summarise(freq=n()) %>% ungroup()) %>% 
  mutate(percentage=(total/freq))

#> Joining, by = "day"
#> # A tibble: 8 x 5
#> # Groups:   day [4]
#>   day   sex    total  freq percentage
#>   <fct> <fct>  <int> <int>      <dbl>
#> 1 Fri   Female     9    19      0.474
#> 2 Fri   Male      10    19      0.526
#> 3 Sat   Female    28    87      0.322
#> 4 Sat   Male      59    87      0.678
#> 5 Sun   Female    18    76      0.237
#> 6 Sun   Male      58    76      0.763
#> 7 Thur  Female    32    62      0.516
#> 8 Thur  Male      30    62      0.484

^{由 reprex 包于 2022 年 4 月 2 日创建（v2.0.1）}

Answer 1

更新

用

..

替换

after_stat

（在 ggplot2 3.4.0 中已弃用）的所有使用，并使用

ave()

代替

tapply()

，可以这样实现：

library(ggplot2)
# For the "tips" dataset
library(reshape2)

ggplot(tips, aes(x = day, group = sex)) +
  geom_bar(aes(
    y = after_stat(prop),
    fill = factor(after_stat(group))
  ), stat = "count") +
  geom_text(aes(
    label = scales::percent(
      after_stat(ave(count, x, FUN = function(x) x / sum(x)))
    ),
    y = after_stat(prop)
  ), stat = "count", position = position_stack(vjust = 0.5)) +
  labs(y = "Percent", fill = "sex") +
  scale_y_continuous(labels = scales::percent)

原答案

使用

after_stat

（而不是

..

）和自定义辅助函数来计算百分比，您可以计算每天的百分比并像这样添加标签。基本上，这涉及计算每个 x 轴组或类别的比例：

library(ggplot2)
library(reshape2)

comp_pct <- function(count, group) {
  count / tapply(count, group, sum)[group]
}

ggplot(tips, aes(x = day, group = sex)) +
  geom_bar(aes(y = ..prop.., fill = factor(..group..)), stat = "count") +
  geom_text(aes(
    label = after_stat(scales::percent(comp_pct(count, x))),
    y = ..prop..
  ), stat = "count", position = position_stack(vjust = 0.5)) +
  labs(y = "Percent", fill = "sex") +
  scale_y_continuous(labels = scales::percent)