ggplot数据筛选：跨数据集绘制密度图用subset还是ifelse？

数据集结构 #

dataset1

Real Wage 1    PPA
1244           105
1577           90
1865           105
1756           105
1634           90
1273           90
2719           105
...            ....

dataset2

Real Wage 2    PPA
1233           105
1588           90
1265           105
1743           105
1224           90
1983           90
2449           105
...            ....

初始可运行代码 #

ggplot() + 
  geom_density( aes( x = dataset1$`Real Wage 1`), fill = "red",  alpha = 0.5)+
  geom_density( aes( x = dataset2$`Real Wage 2`), fill = "blue", alpha = 0.5)+
  theme_classic()

问题场景 #

需要按PPA == 105筛选两个数据集的工资变量后绘图，尝试以下代码但无效：

ggplot() + 
  geom_density( aes( x = subset(dataset1$`Real Wage 1`, PPA == 105)), fill = "red",  alpha = 0.5)+
  geom_density( aes( x = subset(dataset2$`Real Wage 2`, PPA ==105)), fill = "blue", alpha = 0.5)+
  theme_classic()

原因是dataset1$Real Wage 1``仅提取了单个列，subset无法找到PPA列作为筛选条件，尝试ifelse也未成功。

解决方案 #

方法1：直接在图层中对数据集做子集筛选 #

先对整个数据集按PPA筛选，再提取工资列：

ggplot() + 
  geom_density(aes(x = subset(dataset1, PPA == 105)$`Real Wage 1`), 
               fill = "red", alpha = 0.5)+
  geom_density(aes(x = subset(dataset2, PPA == 105)$`Real Wage 2`), 
               fill = "blue", alpha = 0.5)+
  theme_classic()

方法2：提前筛选数据（代码更清晰） #

先将筛选后的数据集存为新对象，再绘图：

# 提前筛选符合条件的数据
dataset1_filtered <- subset(dataset1, PPA == 105)
dataset2_filtered <- subset(dataset2, PPA == 105)

# 绘制密度图
ggplot() + 
  geom_density(aes(x = dataset1_filtered$`Real Wage 1`), 
               fill = "red", alpha = 0.5)+
  geom_density(aes(x = dataset2_filtered$`Real Wage 2`), 
               fill = "blue", alpha = 0.5)+
  theme_classic()

方法3：合并数据集后分组绘图（ggplot推荐方式） #

将两个数据集整理为整洁格式，用分组变量区分来源，自动生成图例，更符合ggplot设计逻辑：

library(dplyr)
library(tidyr)

# 整理为tidy格式数据
tidy_data <- bind_rows(
  dataset1 %>% 
    filter(PPA == 105) %>% 
    select(wage = `Real Wage 1`) %>% 
    mutate(source = "dataset1"),
  dataset2 %>% 
    filter(PPA == 105) %>% 
    select(wage = `Real Wage 2`) %>% 
    mutate(source = "dataset2")
)

# 绘图
ggplot(tidy_data, aes(x = wage, fill = source)) +
  geom_density(alpha = 0.5) +
  scale_fill_manual(values = c("dataset1" = "red", "dataset2" = "blue")) +
  theme_classic()

内容的提问来源于stack exchange，提问作者io_boh