如何在facet_wrap布局下为多个geom_vline添加匹配颜色的图例

问题根源 #

你对ggplot2的美学映射逻辑使用有误：aes()内部传入的color参数是映射到分类维度的标签值，不是直接设置的颜色值。你之前在aes里写color="red"相当于生成了一个值为"red"的分类，和scale_color_manual里定义的median、p 80、p 20三个分类完全不匹配，因此找不到对应颜色，显示默认灰色；如果直接在geom_vline参数里硬编码color值，不属于美学映射范畴，ggplot2不会将这类元素加入图例，因此图例不显示。

修复后代码 #

直接将aes中的color值改为和scale_color_manual对应的分类标签即可：

ggp +
  geom_vline(aes(xintercept = median(overall_vector), color = "median"), linetype = "dashed") +
  geom_vline(aes(xintercept = quantile(overall_vector,0.80), color = "p 80"), linetype = "dashed") + 
  geom_vline(aes(xintercept = quantile(overall_vector,0.20), color = "p 20"), linetype = "dashed") +
  scale_color_manual(name = "statistics", values = c(median = "red", "p 80" = "green", "p 20" = "orange"))

更易维护的推荐写法 #

可以先将所有统计值整理为数据框，仅用一行geom_vline即可完成绘制，后续调整统计项也更方便，且完全兼容facet_wrap布局：

# 构造统计值数据框
stat_df <- data.frame(
  stat_name = c("median", "p 80", "p 20"),
  stat_value = c(median(overall_vector), quantile(overall_vector, 0.8), quantile(overall_vector, 0.2))
)

# 绘图
ggp +
  geom_vline(data = stat_df, aes(xintercept = stat_value, color = stat_name), linetype = "dashed") +
  scale_color_manual(name = "statistics", values = c(median = "red", "p 80" = "green", "p 20" = "orange"))

内容的提问来源于stack exchange，提问作者Xavier