Search code examples
rplotpcacumsumr-corrplot

Adding additional information under corrplot's x axis

I'm trying to add "cumulative proportional variance explained" (from PCA) under corrplot's x axis. I referenced corrplot manual but didn't find out any instruction for doing that. Below is the code I have at the moment using example data.

library("FactoMineR")
library("factoextra")
library("corrplot")

data(decathlon2)
decathlon2.active <- decathlon2[1:23, 1:10]

res.pca <- PCA(decathlon2.active, graph = FALSE)
var <- get_pca_var(res.pca)
corrplot(var$cos2, is.corr=FALSE)

##getting cumulative variance explained from res.pca
variance <- res.pca$eig*100/sum(res.pca$eig)
cumvar <- cumsum(variance)

The problem is how to insert the cumvar information into the corrplot's x-axis, such that they match with the corresponding dim* on top of thecorrplot, which obviates the need to do a scree plot.

Does anyone know how to do that? Any help will be appreciated.

Solution

  • If you want to display these numbers at the bottom of x-axis:

    1. Repeating your calculations

      library("FactoMineR")
library("factoextra")
library("corrplot")

data(decathlon2)
decathlon2.active <- decathlon2[1:23, 1:10]

res.pca <- PCA(decathlon2.active, graph = FALSE)
var <- get_pca_var(res.pca)

variance <- res.pca$eig*100/sum(res.pca$eig)
cumvar <- cumsum(variance)

    2. Making a plot with extra space at the bottom

      corrplot(var$cos2, is.corr=FALSE, mar=c(4,0,0,0))

    3. Adding proportion of variance explained under the cells:

      text(1:5, 0, round(cumvar[1:5], 2), xpd=TRUE)

    And the result:

    result