R corrplot：Plot 相关系数以及显着性星？

Question

Using R corrplot, I have not found a solution where the correlation coefficients in the boxes are plotted together with their significances, ie 0.84*** Here is the code plotting only the significance stars. How can the correlation coefficients be added there?

M<-cor(mtcars)
res1 <- cor.mtest(mtcars, conf.level = .95)
corrplot(cor(mtcars),
     method="square",
     type="lower",
     p.mat = res1$p,
     insig = "label_sig",
     sig.level = c(.001, .01, .05),
     pch.cex = 0.8,
     pch.col = "red",
     tl.col="black",
     tl.cex=1,
     outline=TRUE)

If I add, as by the first answer suggested, addCoef.col = "black", the text overlays the significance stars so they cannot really be seen anymore:

Answer 1

You only have to add the option addCoef.col = "black" to corrplot

Answer 2

To fix the colour clash of the stars I adapted this approach to have bigger stars with a white colour. I use this with the following colours and layout. I can't avoid the overlap but it is legible enough for me to work with:

cex.before <- par("cex")
par(cex = 0.7)
col <- colorRampPalette(c("#BB4444", "#EE9988", "#FFFFFF", "#77AADD", "#4477AA"))
corrplot::corrplot(cor(mtcars), 
    method="color", 
    col=col(200),  
    type="lower", 
    # Combine with significance
    p.mat = res1$p, 
    insig = "label_sig",
    sig.level = c(.001, .01, .05), 
    pch.cex = 3, # Increase size of stars
    pch.col = "white", # Colour of stars
    # hide correlation coefficient on the principal diagonal
    diag=FALSE,
    addCoef.col = "black", # Add coefficient of correlation
    tl.col="black", tl.srt=45, #Text label color and rotation
    tl.cex = 1/par("cex"), cl.cex = 1/par("cex") #Reduce text size of coefficients               
)
par(cex = cex.before)

Answer 3

The position of the significance stars is defined by the place_points function within the corrplot function.

Problem:

If both, the correlation coefficients and the significance level should be displayed, they overlap (I took yellow for the stars since I have some issues with colour vision...).

library(corrplot)
#> corrplot 0.90 loaded

M<-cor(mtcars)
res1 <- cor.mtest(mtcars, conf.level = .95)

corrplot(cor(mtcars),
         method="square",
         type="lower",
         p.mat = res1$p,
         insig = "label_sig",
         sig.level = c(.001, .01, .05),
         pch.cex = 0.8,
         pch.col = "yellow",
         tl.col="black",
         tl.cex=1,
         addCoef.col = "black",
         tl.pos="n",
         outline=TRUE)

^{Created on 2021-10-13 by the reprex package (v2.0.1)}

Quick and temporary (you have to re-do this step everytime you newly loaded the corrplot package) solution:

Change the place_points function within the corrplot function. To do so, run:

trace(corrplot, edit=TRUE)

Then replace on line 443

place_points = function(sig.locs, point) {
  text(pos.pNew[, 1][sig.locs], pos.pNew[, 2][sig.locs], 
       labels = point, col = pch.col, cex = pch.cex, 
       lwd = 2)

with:

# adjust text(X,Y ...) according to your needs, here +0.25 is added to the Y-position    
place_points = function(sig.locs, point) {
      text(pos.pNew[, 1][sig.locs], (pos.pNew[, 2][sig.locs])+0.25, 
           labels = point, col = pch.col, cex = pch.cex, 
           lwd = 2)

and then hit the "Save" button.

Result:

library(corrplot)
#> corrplot 0.90 loaded

#change the corrplot function as described above 
trace(corrplot, edit=TRUE)
#> Tracing function "corrplot" in package "corrplot"
#> [1] "corrplot"

M<-cor(mtcars)
res1 <- cor.mtest(mtcars, conf.level = .95)

corrplot(cor(mtcars),
         method="square",
         type="lower",
         p.mat = res1$p,
         insig = "label_sig",
         sig.level = c(.001, .01, .05),
         pch.cex = 0.8,
         pch.col = "yellow",
         tl.col="black",
         tl.cex=1,
         addCoef.col = "black",
         tl.pos="n",
         outline=TRUE)

^{Created on 2021-10-13 by the reprex package (v2.0.1)}