使用ggplot创建多个变量的组合条形图

Question

我正在尝试使用以下数据创建频率（在% terms中）条形图：

>fulldata
Type Category
Sal         0
Sal         0
Sal         1
Sal         0
Sal         1
Sal         1
Self        1
Self        0
Self        1
Self        0
Self        0

因此，我试图创建一个条形图（使用ggplot），该条形图同时显示完整数据中Sal和Self的百分比以及Sal中Self和Category==1的百分比（带有％值的标签）。我尝试通过从完整数据中过滤Category==1来创建单独的数据帧，但是它们彼此重叠。我尝试了以下方法：

> Category1 = fulldata[which(fulldata$Category==1),]

ggplot(fulldata, aes(x=Type,y = (..count..)/sum(..count..)))+
    geom_bar()+
    geom_label(stat = "count", aes(label=round(..count../sum(..count..),3)*100), 
               vjust=1.2,size=3, format_string='{:.1f}%')+
    scale_y_continuous(labels = scales::percent)+
    labs(x = "Type", y="Percentage")+
    geom_bar(data = Category1, position = "dodge", color = "red")

*原始数据大约有80000行。

Answer 1

一种可能的解决方案是从计算ggplot2中的所有比例开始。

这里是一个假的例子：

df <- data.frame(Type = sample(c("Sal","Self"),100, replace = TRUE),
                 Category = sample(c(0,1),100, replace = TRUE))

我们可以如下计算每个比例以获得最终数据帧：
library(tidyr)
library(dplyr)

df %>% group_by(Category, Type) %>% count() %>% 
  pivot_wider(names_from = Category, values_from = n) %>%
  mutate(Total = `0`+ `1`) %>%
  pivot_longer(-Type, names_to = "Category", values_to = "n") %>%
  group_by(Category) %>%
  mutate(Percent = n / sum(n))

# A tibble: 6 x 4
# Groups:   Category [3]
  Type  Category     n Percent
  <fct> <chr>    <int>   <dbl>
1 Sal   0           27   0.458
2 Sal   1           22   0.537
3 Sal   Total       49   0.49 
4 Self  0           32   0.542
5 Self  1           19   0.463
6 Self  Total       51   0.51 

然后，如果您具有到ggplot2的序列，则可以在一个序列中获得barg raph：
df %>% group_by(Category, Type) %>% count() %>% 
  pivot_wider(names_from = Category, values_from = n) %>%
  mutate(Total = `0`+ `1`) %>%
  pivot_longer(-Type, names_to = "Category", values_to = "n") %>%
  group_by(Category) %>%
  mutate(Percent = n / sum(n)) %>%
  ggplot(aes(x = reorder(Category, desc(Category)), y = Percent, fill = Type))+
  geom_col()+
  geom_text(aes(label = scales::percent(Percent)), position = position_stack(0.5))+
  scale_y_continuous(labels = scales::percent)+
  labs(y = "Percentage", x = "Category")

它回答了您的问题吗？