R：如何可视化连续数据和分类数据之间的关系

Question

我有以下data.frame，其中包含3个分类变量（不同类型的血管病变）和1个连续变量（输出）。我有兴趣看到输出与不同类型的血管病变之间的关系，即与轻度/重度病理学相关的更高/更低的输出？

> dput(df)
structure(list(Vascular_Pathology_M = structure(c(1L, 2L, 3L, 
1L, 1L, 2L, 4L, 3L, 1L, 2L), .Label = c("Absent", "Mild", "Mild/Moderate", 
"Moderate/Severe", "Severe"), class = "factor"), Vascular_Pathology_F = structure(c(4L, 
2L, 1L, 1L, 1L, 1L, 2L, 4L, 1L, 1L), .Label = c("Absent", "Mild", 
"Mild/Moderate", "Moderate/Severe", "Severe"), class = "factor"), 
    Vascular_Pathology_O = structure(c(1L, 3L, 4L, 3L, 1L, 2L, 
    1L, 1L, 1L, 2L), .Label = c("Absent", "Mild", "Mild/Moderate", 
    "Moderate/Severe"), class = "factor"), Output = c(1.01789418758932, 
    1.05627630598801, 1.49233946102323, 1.38192374975672, 1.13097652937671, 
    0.861306979571144, 0.707820561413699, 1.16628243128399, 0.983163398006992, 
    1.23972603843843)), .Names = c("Vascular_Pathology_M", "Vascular_Pathology_F", 
"Vascular_Pathology_O", "Output"), row.names = c(1L, 3L, 4L, 
5L, 6L, 7L, 8L, 10L, 11L, 12L), class = "data.frame")
> df
   Vascular_Pathology_M Vascular_Pathology_F Vascular_Pathology_O    Output
1                Absent      Moderate/Severe               Absent 1.0178942
3                  Mild                 Mild        Mild/Moderate 1.0562763
4         Mild/Moderate               Absent      Moderate/Severe 1.4923395
5                Absent               Absent        Mild/Moderate 1.3819237
6                Absent               Absent               Absent 1.1309765
7                  Mild               Absent                 Mild 0.8613070
8       Moderate/Severe                 Mild               Absent 0.7078206
10        Mild/Moderate      Moderate/Severe               Absent 1.1662824
11               Absent               Absent               Absent 0.9831634
12                 Mild               Absent                 Mild 1.2397260

Answer 1

您可以查看各种病症的相互作用。例如，使用条形图

## Make the interaction variable
df$interact <- interaction(df[, 1:3], sep="_")

## Look at means of groups
library(dplyr)
df %>% group_by(interact) %>%
  dplyr::summarise(Output = mean(Output)) -> means

ggplot(means, aes(interact, Output))+
  geom_bar(stat="identity") +
  theme(axis.text=element_text(angle=90)) +
  xlab("Interaction")

或点

ggplot(df, aes(interact, Output))+
  geom_point() +
  theme(axis.text=element_text(angle=45, hjust=1)) +
  xlab("Interaction") +
  geom_point(data=means, col="red") +
  ylim(0, 1.6)

Answer 2

您可以根据分类变量

简单地绘制输出

plot(df[, 1], df[, 4])
plot(df[, 2], df[, 4])
plot(df[, 3], df[, 4])

Answer 3

您有一个4维数据集。一个选项是在一个小的多个系列中进行散点图（x / y =两个维度）（还有一个维度），并将Output变量映射到视觉上像大小一样的东西（那里是第四个）尺寸）。

示例，将数据放入名为data.frame的{​​{1}}后（因为my_dat已分配给R中的某个函数）。抖动点以显示每个点的多个观察值，并按Y位置着色以帮助明确哪个点与哪个类别相关。

df