scale_*_discrete为何不显示全部标签？如何配置含缺失项坐标轴

需要绘制存在缺失值的数据（如周期表某周期部分原子数据缺失），要求x轴保留缺失原子的位置标签但不显示对应(x,y)点。当前使用scale_x_discrete时，lims中的值未全部显示，需解决以下问题：

1. 如何配置才能显示所有缺失原子的标签？
2. 能否让lims中的值按实际间距排列？
3. scale_x_discrete是否为合适工具？

附原测试代码：

vals <- seq(1.1:11.1)
shortlist <- c("Ce","Pr","Nd","Sm","Gd","Tb","Dy","Ho","Er","Tm","Yb")
radii <- c(1.010,0.990,0.983,0.958,0.938,0.923,0.912,0.901,0.890,0.880,0.868)
df<-do.call(rbind, Map(data.frame, shortlist=shortlist,radii=radii,vals=vals))

#lists for labels with extra atoms
lims <- c(1.032,1.010,0.990,0.983,0.97,0.958,0.947,0.938,0.923,0.912,0.901,0.890,0.880,0.868,0.861)
labs<-c("La","Ce","Pr","Nd","Pm","Sm","Eu","Gd","Tb","Dy","Ho","Er","Tm","Yb","Lu")

#Why does it not include all the values in 'lims'?
ggplot(data=df) + 
   geom_point(aes(x=factor(radii), y=vals)) + 
   scale_x_discrete(breaks = lims, labels=labs)

1. 原代码的问题根源 #

你将连续型的半径值用factor(radii)转成了离散因子，而scale_x_discrete的breaks仅能识别数据中存在的因子水平——缺失原子对应的半径值不在原始radii列表里，因此无法被显示。

2. 显示所有缺失原子标签的方法 #

不能直接用scale_x_discrete实现，需先构建包含**所有原子（含缺失项）**的基准数据框，再将现有数据匹配到该框架中：

• 创建包含全量原子名称、对应半径的完整数据框，缺失原子的vals设为NA
• 基于完整数据框绘图，geom_point会自动忽略NA值，不显示对应点

3. 按实际半径间距排列的实现 #

若要按半径的真实数值间距排列，x轴需使用连续型变量（半径值），而非离散的原子名称。通过scale_x_continuous设置刻度和标签，即可让刻度按半径的实际大小间距分布。

两种可行代码示例 #

示例一：x轴显示原子名称（按原子顺序离散排列） #

library(ggplot2)

# 构建全量原子的基准数据框
full_labs <- c("La","Ce","Pr","Nd","Pm","Sm","Eu","Gd","Tb","Dy","Ho","Er","Tm","Yb","Lu")
full_radii <- c(1.032,1.010,0.990,0.983,0.97,0.958,0.947,0.938,0.923,0.912,0.901,0.890,0.880,0.868,0.861)
full_df <- data.frame(shortlist = full_labs, radii = full_radii)

# 将现有数据的vals匹配到全量数据框，缺失项自动为NA
full_df$vals <- df$vals[match(full_df$shortlist, df$shortlist)]

# 绘图：自动隐藏缺失点，保留所有原子标签
ggplot(full_df, aes(x = shortlist, y = vals)) +
  geom_point() +
  xlab("原子") + ylab("数值")

示例二：x轴按半径实际间距排列（连续型，显示原子名称标签） #

library(ggplot2)

# 直接用原始数据绘图，x轴为连续型半径值
ggplot(df, aes(x = radii, y = vals)) +
  geom_point() +
  # 设置全量半径为刻度点，对应标签为原子名称
  scale_x_continuous(breaks = full_radii, labels = full_labs, limits = range(full_radii)) +
  xlab("原子半径") + ylab("数值")

总结 #

• scale_x_discrete适合x轴为分类变量的场景，若需要按连续数值间距排列，应使用scale_x_continuous
• 要保留缺失标签，必须让x轴的基准包含所有需要显示的类别/数值，再将现有数据匹配进去

内容的提问来源于stack exchange，提问作者S B