R-在x和y轴上使用具有不同变量的corrplot

Question

As you all probably know, corrplot can be used to create beautiful plots that visualize the strength of the relationship amongst a set of variables, with the same variables on the x-axis as on the y-axis (usually symmetrical, unless you have a different metric on the upper diagonal than the lower diagonal). I want to use corrplot in a slightly different way.

Suppose I have a dataframe that look like this:

var1 var2 beta  se   pvalue
X    a    .01   .01  .35
X    b    -.02  .02  .45
X    c    .04   .01  .55
X    d    .04   .01  .55
Y    a    .06   .01  .01
Y    b    -.02  .02  .25
Y    c    .04   .01  .55
Y    d    .02   .01  .55
Z    a    .04   .01  .01
Z    b    -.01  .01  .45
Z    c    .03   .02  .02
Z    d    .04   .02  .02

Can I use corrplot to create a figure with variables X,Y, and Z on the x-axis and a, b, c, and d on the Y-axis? If yes, how?

PS. The value I want to plot is beta.

EDIT: I edited the example file to be non-symmetric with respect to the nr of variables on x and y

Answer 1

You can reshape your data into a square matrix and just feed it into corrplot . This works even if the resulting matrix is not square.

## Your new data
Dat = read.table(text="var1 var2 beta  se   pvalue
X    a    .01   .01  .35
X    b    -.02  .02  .45
X    c    .04   .01  .55
X    d    .04   .01  .55
Y    a    .06   .01  .01
Y    b    -.02  .02  .25
Y    c    .04   .01  .55
Y    d    .02   .01  .55
Z    a    .04   .01  .01
Z    b    -.01  .01  .45
Z    c    .03   .02  .02
Z    d    .04   .02  .02",
header=TRUE)

## Now reshape the data  (same as before)
wide = reshape(Dat[,1:3], idvar = c("var1"),
    timevar="var2", direction = "wide")
rownames(wide) = wide$var1
wide = wide[,-1]
colnames(wide) = sub("beta.", "", colnames(wide))

## Pass it to corrplot
library(corrplot)
corrplot(as.matrix(wide), is.corr=FALSE, tl.srt=0)