使用ggrepel失败时，如何解决ggplot柱状图标签重叠问题？

看起来你遇到的问题核心是临时计算的..prop..变量无法被ggrepel直接识别，以及stat参数不匹配导致的报错。咱们一步步来解决：

错误原因分析 #

• "object 'prop' not found"：..prop..是ggplot绘制geom_bar时内部临时生成的比例值，它并没有存在于你的原始数据框df中。直接在geom_text_repel里调用y = ..prop..，ggrepel找不到这个变量，自然会报错。
• "stat_count requires the following missing aesthetics: x"：当你把ggrepel参数放进全局aes()时，默认stat是count，但没有正确传递x轴的映射逻辑，和柱状图的stat规则冲突，就会出现缺失x参数的提示。

最优解决方案：提前计算比例（推荐） #

最稳妥的方式是先在数据预处理阶段计算好每个分组的比例，这样后续绘图和添加标签时完全不依赖ggplot的临时变量，逻辑更清晰，也能彻底避免报错。

步骤1：预处理数据计算比例 #

用dplyr按种族和就业状态分组，算出每组的占比：

library(dplyr)
library(ggplot2)
library(ggrepel)

# 预处理：计算每个种族在各就业状态下的比例
df_processed <- df %>%
  filter(!is.na(`Self Reported Race (roll up)_Cleaned`)) %>%
  group_by(`Self Reported Race (roll up)_Cleaned`, C19_Employment) %>%
  summarise(n = n(), .groups = "drop") %>%
  group_by(`Self Reported Race (roll up)_Cleaned`) %>%
  mutate(prop = n / sum(n)) %>%
  ungroup()

步骤2：绘制带ggrepel标签的柱状图 #

用预处理好的数据绘图，直接调用计算好的prop列，添加geom_text_repel时对齐柱状图位置：

ggplot(df_processed, aes(
  x = C19_Employment,
  y = prop,
  fill = `Self Reported Race (roll up)_Cleaned`,
  group = `Self Reported Race (roll up)_Cleaned`
)) +
  geom_bar(stat = "identity", position = "dodge", na.rm = TRUE) +
  # 添加ggrepel标签，匹配柱状图的分组间距
  geom_text_repel(
    aes(label = scales::percent(prop, accuracy = 1)), # 格式化为百分比
    position = position_dodge(width = 0.9), # 和柱状图默认dodge宽度一致
    vjust = -0.5, # 标签放在柱子上方
    size = 3.5
  ) +
  labs(
    title = "Employment Status by Self-Reported Race",
    x = "Employment Status",
    y = "Proportion of Race",
    fill = "Self-Reported Race"
  ) +
  scale_y_continuous(labels = scales::percent) + # y轴显示百分比
  theme(legend.position = "bottom")

另一种方案：在ggplot内直接计算比例（无需提前预处理） #

如果你不想提前处理数据，也可以给geom_text_repel指定stat = "count"，让它和geom_bar用同样的统计逻辑计算比例：

df %>%
  filter(!is.na(`Self Reported Race (roll up)_Cleaned`)) %>%
  ggplot(aes(
    x = C19_Employment,
    fill = `Self Reported Race (roll up)_Cleaned`,
    group = `Self Reported Race (roll up)_Cleaned`
  )) +
  geom_bar(position = "dodge", na.rm = TRUE) +
  geom_text_repel(
    aes(
      y = ..prop..,
      label = scales::percent(..prop.., accuracy = 1)
    ),
    stat = "count", # 和柱状图用相同的统计规则
    position = position_dodge(width = 0.9),
    vjust = -0.5,
    size = 3.5,
    na.rm = TRUE
  ) +
  labs(
    title = "Employment Status by Self-Reported Race",
    x = "Employment Status",
    y = "Proportion of Race",
    fill = "Self-Reported Race"
  ) +
  scale_y_continuous(labels = scales::percent) +
  theme(legend.position = "bottom")

额外小提示 #

• 用scales::percent()能把比例转成更易读的百分比标签，避免显示小数。
• 如果标签还是拥挤，可以调整size（字体大小）、force（ggrepel的排斥力度）或者hjust（水平位置）来优化。

内容的提问来源于stack exchange，提问作者jhalula