计算 geom_bar 中分组值的百分比值，同时显示每个分组值的计数

Question

我正在尝试使用 ggplot2 在堆积栏中显示计数和百分比数据。数据在这里

https://github.com/oiramsch/CvsRWO/blob/master/colorList.csv

我试过了：

p <- colorList %>% ggplot(mapping = aes(x=factor(age, levels = c("young","old")),fill=factor(answer_correct, levels = c(0,1), labels = c("False","Correct"))))

然后

p + geom_bar(stat="count", position="stack")+facet_grid(~condition)+labs(title="Colors",subtitle = "per setsize grouped by age", x="", y="Count answer correct", fill="Answer correct")+geom_text(aes(label = paste(..count..,"\n",scales::percent(..count../1298,0.01))),position=position_stack(0.5),stat="count")

结果我收到了：

问题是我使用固定值 1298 作为总值来计算百分比份额（不幸的是，并非所有列都如此）。所以我正在寻找一个 ggplot 的内部变量，我可以使用它来访问每列的总和，甚至更好地访问每个集合大小的错误/正确值的百分比。但我无法弄清楚如何做到这一点。当然，我可以提前总结数据，但我认为 ggplot 内部必须有一个解决方案......如果有人知道如何修复我的代码，我将不胜感激。

Answer 1

我建议采用这种方法总结价值，这可能是您的一个选择：

library(ggplot2)
library(dplyr)
#Data
colorList <- read.csv('https://raw.githubusercontent.com/oiramsch/CvsRWO/master/colorList.csv',stringsAsFactors = F)
#Aggregate
colorList %>% mutate(age=factor(age, levels = c("young","old")),
             answer_correct=factor(answer_correct, levels = c(0,1))) %>%
  group_by(age,condition,answer_correct) %>% summarise(N=n()) %>% ungroup() %>%
  group_by(age,condition) %>% 
  mutate(Total=sum(N),Percent=N/Total,
         Lab=paste0(N,' (',paste0(round(100*Percent,0),'%'),')')) -> Sums
#Plot
ggplot(Sums,aes(x=age,y=N,fill=answer_correct))+
  geom_bar(stat='identity',position = position_stack())+
  facet_wrap(.~condition,scales = 'free')+
  geom_text(aes(label=Lab),position = position_stack(vjust = .5),size=3)+
  geom_text(aes(y=Total,label=Total),vjust=-0.25,size=3)

输出：

Answer 2

你在找(..count..)/sum(..count..)吗？

p + 
  geom_bar(stat="count", position="stack")+
  facet_grid(~condition)+
  labs(title="Colors",subtitle = "per setsize grouped by age", x="", y="Count answer correct", fill="Answer correct")+
  geom_text(aes(label = paste(..count..,"\n",
                              scales::percent(..count../sum(..count..), 0.01))),
            position=position_stack(0.5),stat="count")