Plot elements of a matrix with different colours for a specific column

Question

I have the following correlation matrix.

       AUS   AUT   CAN   CHE   DEU  EU15   FRA   GBR   ITA   JPN   USA
AUS  1.000 0.058 0.476 0.313 0.111 0.277 0.184 0.296 0.202 0.192 0.267
AUT  0.058 1.000 0.254 0.658 0.749 0.761 0.626 0.387 0.460 0.410 0.278
CAN  0.476 0.254 1.000 0.390 0.321 0.534 0.377 0.538 0.391 0.231 0.746
CHE  0.313 0.658 0.390 1.000 0.604 0.706 0.610 0.310 0.565 0.437 0.305
DEU  0.111 0.749 0.321 0.604 1.000 0.859 0.620 0.387 0.472 0.520 0.369
EU15 0.277 0.761 0.534 0.706 0.859 1.000 0.808 0.682 0.713 0.601 0.531
FRA  0.184 0.626 0.377 0.610 0.620 0.808 1.000 0.467 0.553 0.444 0.357
GBR  0.296 0.387 0.538 0.310 0.387 0.682 0.467 1.000 0.324 0.407 0.591
ITA  0.202 0.460 0.391 0.565 0.472 0.713 0.553 0.324 1.000 0.492 0.315
JPN  0.192 0.410 0.231 0.437 0.520 0.601 0.444 0.407 0.492 1.000 0.321
USA  0.267 0.278 0.746 0.305 0.369 0.531 0.357 0.591 0.315 0.321 1.000

My goal: plot all correlations in ascending order assigning different colours to the correlations of USA.

I have accomplished to plot all the correlations like this:

x[lower.tri(x)] <- NA
diag(x) <- NA
x <- as.vector(gdpcor)
x <- x[!is.na(x)]
x <- x[order(x)]
plot(x)

But I can't figure out how to assign a different colour for USA's correlations. Any ideas?

Answer 1

How about starting with something like:

x <- as.matrix(read.table(text="AUS   AUT   CAN   CHE   DEU  EU15   FRA   GBR   ITA   JPN   USA
AUS  1.000 0.058 0.476 0.313 0.111 0.277 0.184 0.296 0.202 0.192 0.267
AUT  0.058 1.000 0.254 0.658 0.749 0.761 0.626 0.387 0.460 0.410 0.278
CAN  0.476 0.254 1.000 0.390 0.321 0.534 0.377 0.538 0.391 0.231 0.746
CHE  0.313 0.658 0.390 1.000 0.604 0.706 0.610 0.310 0.565 0.437 0.305
DEU  0.111 0.749 0.321 0.604 1.000 0.859 0.620 0.387 0.472 0.520 0.369
EU15 0.277 0.761 0.534 0.706 0.859 1.000 0.808 0.682 0.713 0.601 0.531
FRA  0.184 0.626 0.377 0.610 0.620 0.808 1.000 0.467 0.553 0.444 0.357
GBR  0.296 0.387 0.538 0.310 0.387 0.682 0.467 1.000 0.324 0.407 0.591
ITA  0.202 0.460 0.391 0.565 0.472 0.713 0.553 0.324 1.000 0.492 0.315
JPN  0.192 0.410 0.231 0.437 0.520 0.601 0.444 0.407 0.492 1.000 0.321
USA  0.267 0.278 0.746 0.305 0.369 0.531 0.357 0.591 0.315 0.321 1.000"))


x[lower.tri(x)] <- NA
diag(x) <- NA

df <- subset(as.data.frame(as.table(x), responseName = 'Corr'),!is.na(Corr))
df <- df[order(df$Corr),]

ggplot(df, aes(x=1:nrow(df),y=Corr,col=Var2=='USA')) + geom_point()

side note: In case you haven't tried it, check out library(corrplot) as a good way to visualize correlations. For example:

corrplot(x, is.corr = FALSE, method='square', diag=FALSE)