Plotting a bar plot where Y axis reflects percents, how can I "complete" each bar to 100%?

在绘制Y轴反映百分比的条形图时，我如何才能将每个条形图“完成”到100%？

Example

Consider the following plot:

考虑以下图：

library(ggplot2)
library(tibble)

my_df <- tribble(~step, ~prop,
                   "a", 1,
                   "b", 0.8,
                   "c", 0.55,
                   "d", 0.2
)

my_df |> 
  ggplot(aes(x = step,
             y = prop)) +
  geom_col()

^{Created on 2023-09-10 with reprex v2.0.2}

创建于2023-09-10，Reprex v2.0.2

How can I create a plot that completes each bar to 100% out of it with dashed lines? Something like:

我如何创建一个用虚线100%地完成每个条形图的绘图？类似于：

One option would be use a second geom_col to first draw 100% bars, then place your bars on top of it where we have to explicitly set the outline color. Note: One downside, which I did not realize at first, is that the outline of the background bars shouldn't but is visible. A fix would be to set the linewidth and or the bar width as suggested by @jkd and @r2evans in their comments. A second and cleaner option would be to use a geom_path (see below).

一种选择是使用第二个geom_ol首先绘制100%的条形图，然后将条形图放置在必须显式设置轮廓颜色的位置。注意：一开始我没有意识到的一个缺点是，背景栏的轮廓不应该但却是可见的。一个解决办法是按照@jkd和@r2evans在他们的评论中建议的那样设置行宽和/或条宽。第二个更简洁的选项是使用geom_path(如下所示)。

library(ggplot2)
library(tibble)

my_df <- tribble(
  ~step, ~prop,
  "a", 1,
  "b", 0.8,
  "c", 0.55,
  "d", 0.2
)

my_df |>
  ggplot(aes(
    x = step,
    y = prop
  )) +
  geom_col(aes(y = 1), color = "red", linetype = "dashed", fill = NA) +
  geom_col(color = "grey35")

IMHO, the cleanest option to achieve the same result would be to use a geom_path which however requires some effort to create an appropriate dataset to draw the paths:

IMHO，要获得相同的结果，最干净的选择是使用geom_path，但这需要一些工作来创建适当的数据集来绘制路径：

library(ggplot2)
library(tibble)
library(dplyr, warn = FALSE)

make_path <- function(y, width = .9) {
  if (y < 1) {
    data.frame(
      x = width / 2 * c(-1, -1, 1, 1),
      y = c(y, 1, 1, y)
    )
  } else {
    data.frame(x = numeric(0), y = numeric(0))
  }
}

df_path <- my_df |>
  mutate(step = factor(step)) |> 
  reframe(make_path(prop), .by = step) |>
  mutate(x = as.numeric(step) + x)

my_df |>
  ggplot(aes(
    x = step,
    y = prop
  )) +
  geom_col(color = "grey35") +
  geom_path(
    data = df_path,
    aes(x = x, y = y, group = step),
    color = "red", linetype = "dashed"
  )

Yet another way with stacked bars and controlling alpha and manual scaling of properties:

另一种方法是堆叠条形图并控制Alpha和手动缩放属性：

library(ggplot2)
library(tibble)

my_df <- tribble(~step, ~prop,
                 "a", 1,
                 "b", 0.8,
                 "c", 0.55,
                 "d", 0.2
)
my_df$type <- 'filled'

my_df2 <- my_df
my_df2$prop <- 1 - my_df2$prop
my_df2$type <- 'empty'

my_df <- rbind(my_df, my_df2)

my_df |> ggplot(aes(x = step, y = prop)) + 
         geom_col(aes(alpha=type, col=type, linetype=type)) +
         scale_linetype_manual(values=c('dashed','solid')) +
         scale_color_manual(values=c('red', 'gray')) +
         scale_alpha_manual(values=c(0, 1)) + theme(legend.position = "none") 

In base R we can use rect.

在基数R中，我们可以使用RECT。

plot(d$prop, xlim=c(.5, 4.5), ylim=0:1, type='n', xaxt='n', ylab='Proportion', xlab='Step')
axis(1, axTicks(1), labels=d$step)
sq <- seq_len(nrow(d))
rect(sq - .3, 0, sq + .3, rep(1, nrow(d)), lty=2, border=2)
rect(sq - .3, 0, sq + .3, d$prop, col=8, border=8, lwd=2)

Data:

数据：

d <- structure(list(step = c("a", "b", "c", "d"), prop = c(1, 0.8, 
0.55, 0.2)), class = "data.frame", row.names = c(NA, -4L))