如何重叠 PCA 图中的不同信息？

Question

我尝试绘制代表两种类别的修改后的 PCA 图。
例如，我想用 1) 基于 Kingdom 的 PCA 椭圆绘制 PCA，2) 用 Class 变量着色的 PCA 样本点。
这是示例数据集。

require(ggplot2)
require(ggfortify)

set.seed(1)
df <- structure(list(Sample = c("cat", "dog", "rabbit", "chicken", "duck", "butterfly", "ladybug", "rose", "lily", "iris", "maple tree", "pinetree", "ginkgo"), 
                     Class = c("mammalia", "mammalia", "mammalia", "bird", "bird", "insect", "insect", "flower", "flower", "flower", "tree", "tree", "tree"), 
                     Kingdom = c("animalia", "animalia", "animalia", "animalia", "animalia", "animalia", "animalia", "plantae", "plantae", "plantae", "plantae", "plantae", "plantae")), 
                class = "data.frame", row.names = c(NA, -13L))
rownames(df)<-df[,1]
df[,1]<-NULL

for(i in 3:20){
  df[,i]<-sample(100, size=nrow(df), replace=TRUE)
}
df[,c(3:20)]<-log(df[,c(3:20)]+1, base=2)

df #I ommitted the value from v13 to v20 for simplicity
              Class  Kingdom       V3       V4       V5       V6       V7       V8       V9      V10      V11      V12
cat        mammalia animalia 6.108524 5.781360 5.087463 5.357552 4.247928 5.614710 5.000000 5.129283 5.321928 6.303781
dog        mammalia animalia 5.321928 6.228819 6.409391 5.491853 4.523562 6.491853 4.169925 3.459432 4.643856 4.523562
rabbit     mammalia animalia 1.000000 3.000000 5.169925 4.700440 6.303781 4.584963 6.209453 1.000000 5.754888 6.149747
chicken        bird animalia 5.129283 6.209453 6.149747 6.149747 6.044394 6.409391 6.459432 5.459432 6.539159 4.857981
duck           bird animalia 6.459432 6.321928 6.228819 5.321928 6.149747 4.906891 6.392317 5.906891 6.442943 5.247928
butterfly    insect animalia 5.459432 6.426265 5.426265 5.700440 6.459432 3.807355 6.507795 4.754888 5.357552 5.954196
ladybug      insect animalia 3.906891 5.247928 5.285402 5.426265 6.149747 4.523562 5.614710 4.000000 6.614710 5.554589
rose         flower  plantae 6.375039 6.491853 4.392317 2.807355 6.247928 6.554589 6.022368 5.882643 6.392317 6.087463
lily         flower  plantae 5.906891 5.247928 4.857981 4.643856 6.357552 4.857981 6.569856 4.906891 6.507795 6.442943
iris         flower  plantae 5.700440 5.129283 4.392317 5.044394 6.658211 5.614710 6.599913 4.643856 5.169925 6.643856
maple tree     tree  plantae 6.614710 6.491853 5.491853 3.906891 3.807355 5.087463 5.930737 5.426265 5.459432 6.169925
pinetree       tree  plantae 6.426265 5.491853 6.459432 1.584963 5.357552 5.523562 5.700440 5.614710 1.000000 6.643856
ginkgo         tree  plantae 4.459432 6.321928 6.149747 5.523562 6.491853 4.459432 6.554589 6.266787 4.906891 5.700440

指导数据集后，我使用自动绘图绘制 PCA 图

pca_df<-prcomp(df[,-c(1,2)], scale.=TRUE)
autoplot(pca_df)
autoplot(pca_df, data=df, colour='Kingdom', frame.type='norm')

在这个情节中，我想根据 Class

重叠颜色

pca_df<-prcomp(df[,-c(1,2)], scale.=TRUE)
autoplot(pca_df)
autoplot(pca_df, data=df, colour='Class')

总而言之，我想使用来自 Kingdom 变量的椭圆表示，并使用 Class 变量着色。如何在一个 PCA 图中组合不同的表示形式？

加试
我尝试使用其他包 (factoextra) 绘制 PCA 图，但它显示相同的结果。

require (factoextra)

fviz_pca_ind(pca_df,
             col.ind=df$Kingdom,
             addEllipses = TRUE,
             col.ind.sup=df$Class,
             repel=TRUE)

Answer 1

您可以为省略号指定 frame.colour，为点指定 colour：

autoplot(pca_df, data=df, colour='Class', frame.type='norm', frame.colour = "Kingdom")

如何重叠 PCA 图中的不同信息？

How to overlapping the different information in the PCA plot?

visualization

r

pca