如何使用ggplot的after_stat控制水平堆叠条形图的标签位置

It looks like the core issue here is that ggplot2 isn't properly recognizing how to group your data for stacking, which throws off the label positioning. Let's break down the problem and fix it with two straightforward approaches:

1. Add a group Aesthetic to Enable Proper Stacking #

Your original code doesn't explicitly tell ggplot2 to group the bars by the names variable. Even though you're setting fill colors directly, position_stack() needs a grouping variable to calculate the correct stacked positions for both bars and labels.

Here's the corrected code:

# Recreate your data frame for clarity
df <- data.frame(
  names = c("v1", "v2"),
  count = c(55L, 17L),
  share = c(0.76, 0.24),
  label = c("76 %", "24 %"),
  color = c("#df91a3", "#A5AA99")
)

p1 <- ggplot(df, aes(y = 1, x = share, group = names)) + 
  geom_col(
    position = position_stack(), 
    color = "black", 
    fill = df$color, 
    orientation = "y"
  ) + 
  scale_fill_identity() + 
  geom_text(
    aes(label = names), 
    position = position_stack(vjust = 0.5),  # Centers labels in each bar segment
    size = 7
  )

Why this works: #

• Adding group = names tells ggplot2 to split the data into two distinct groups (v1 and v2) for stacking.
• position_stack(vjust = 0.5) ensures labels align perfectly to the horizontal center of each stacked bar segment, rather than sticking to the edge.

2. Calculate Label Positions Manually with after_stat() #

If you prefer to avoid relying on position_stack() for label placement, you can compute the center position of each bar segment directly using after_stat(). This method uses cumulative sums to find the midpoint of each stacked segment:

p1 <- ggplot(df, aes(y = 1, x = share)) + 
  geom_col(
    position = position_stack(), 
    color = "black", 
    fill = df$color, 
    orientation = "y"
  ) + 
  scale_fill_identity() + 
  geom_text(
    aes(
      x = after_stat(cumsum(x) - x/2),  # Compute midpoint of each stacked segment
      label = names
    ),
    size = 7
  )

Why this works: #

• cumsum(x) gives the cumulative total of share values up to each segment (0.76 for v1, 1.00 for v2).
• Subtracting x/2 adjusts this cumulative sum to the exact midpoint of each segment (0.38 for v1, 0.88 for v2), which is where we want the labels to sit.

Verifying the Fix #

After applying either fix, running ggplot_build(p1) will show the geom_text data now has correct x positions:

• For v1: ~0.38 (center of the 0-0.76 segment)
• For v2: ~0.88 (center of the 0.76-1.00 segment)

Either approach will correctly stack your labels alongside the horizontal bars without needing to precompute positions outside of ggplot2.

内容的提问来源于stack exchange，提问作者Lenn