将合并Corrplot与两组数据进行比较？

Question

I wonder if I could generate a combined corrplot for two groups of data? That is to say, the upper triangular for one group, and the lower triangular for another group. I give an example as follows:

library(corrplot)

mydata <- replicate(5, rnorm(20))

colnames(mydata) <- c('x1','x2','x3','x4','x5')

mydata <- transform(mydata, group = c(rep('A',10),rep('B',10)))

corrplot(cor(mydata[which(mydata$group=='A'),c(1:5)]), method="number", type="upper", title="Group A", mar=c(1,0,1,0))
corrplot(cor(mydata[which(mydata$group=='B'),c(1:5)]), method="number", type="lower", title="Group B", mar=c(1,0,1,0))

Now in this example, the figures are as follows:

It would be interesting to me if somebody could combine these two figures into one figure ? Thanks in advance!

Answer 1

You could replace the values of one correlation matrix with the values from the other.

cor_A <- cor(mydata[which(mydata$group=='A'),c(1:5)])
cor_B <- cor(mydata[which(mydata$group=='B'),c(1:5)])

# replace the values
cor_A[lower.tri(cor_A)] <- cor_B[lower.tri(cor_B)]

corrplot(cor_A, method="number", title="Group A & B", mar=c(1,0,1,0))

data

set.seed(1)
mydata <- replicate(5, rnorm(20))
colnames(mydata) <- c('x1','x2','x3','x4','x5')
mydata <- transform(mydata, group = c(rep('A',10),rep('B',10)))

Answer 2

Here is one possibility:

library(corrplot)    
set.seed(1)
mydata <- replicate(5, rnorm(20))   
colnames(mydata) <- c('x1','x2','x3','x4','x5')    
mydata <- transform(mydata, group = c(rep('A',10),rep('B',10)))

cor_A = cor(mydata[which(mydata$group=='A'),c(1:5)])
cor_B = cor(mydata[which(mydata$group=='B'),c(1:5)])
my_cor = upper.tri(cor_A)*cor_A + lower.tri(cor_B) * cor_B

corrplot(cor(mydata[which(mydata$group=='A'),c(1:5)]), method="number", type="upper", title="Group A", mar=c(1,0,1,0))
corrplot(cor(mydata[which(mydata$group=='B'),c(1:5)]), method="number", type="lower", title="Group B", mar=c(1,0,1,0))
corrplot(my_cor, method="number", title="Group A (upper) and B (lower)", mar=c(1,0,1,0))

---