R PCA : With the fviz_pca_ind function, can we have two categorical variables: one point shape and one fill color?

Question

I am trying to make a PCA plot with individuals -where one categorical variable (A) would be represented as the point shape (eg one group as a circle, a second one as a square, etc.) -and a second categorical variable (B) as the color inside the point Is that possible? Which code would you use?

Please, could you provide your code. – Earl Mascetti Apr 11 '20 at 11:42 — Earl Mascetti, Apr 11 '20 at 11:42

score 6 · Accepted Answer · answered Apr 11 '20 at 13:43

I don't think you can modify the output from fviz_pca_ind(), so you would need to take out the data from the results, and plot it again using ggplot2:

library(factoextra)
library(ggplot2)

data <- iris
colnames(data)[5] <- "A"
data$B <- sample(letters[1:2],nrow(data),replace=TRUE)

res.pca <- prcomp(data[,1:4],  scale = TRUE)
basic_plot <- fviz_pca_ind(res.pca, label="none")

ggplot(cbind(basic_plot$data,data[,c("A","B")]),
aes(x=x,y=y,col=A,shape=B)) + geom_point() + theme_bw()

thanks a lot for this clear answer ! This is exactly what I was looking for :-) — Xav64, Apr 11 '20 at 17:23

R PCA : With the fviz_pca_ind function, can we have two categorical variables: one point shape and one fill color?

1 Answers1