create matrix based on apply function in R-CodePudding

I have a matrix in R like this

dat <- matrix(c(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1,
                0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1,
                0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1,
                0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0, 1), ncol = 4)

and i want to calculate the local variant (lv) of mutual information. For a pair of columns I can use the following code to find the lv

dat <- dat[,c(1,2)]
#install.packages("rinform")
library(rinform)
library(devtools)
#install_github("ELIFE-ASU/rinform")
#install_github("ELIFE-ASU/rinform", ref = "dev")
library(rinform)
mutual_info(dat)
mi <- mutual_info(dat, local = T)
mi <- as.matrix(mi)

which results in:

> mi
            [,1]
 [1,] -1.0000000
 [2,] -1.0000000
 [3,]  0.2223924
 [4,]  0.2223924
 [5,]  0.2223924
 [6,]  0.2223924
 [7,]  0.2223924
 [8,]  0.2223924
 [9,]  0.2223924
[10,]  0.2223924
[11,]  0.2223924
[12,]  0.2223924
[13,]  0.2223924
[14,]  0.2223924
[15,]  0.2223924
[16,]  0.2223924
[17,]  1.5849625
[18,]  1.5849625
[19,]  1.5849625
[20,] -1.5849625

How can I create a matrix that each column will have the mi values of each pair? Tried to use an apply function but couldn't make it work. The output result must be a 20x6 matrix with all pair-wise column combinations, i.e., the above mi procedure for dat[,c(1,2)], dat[,c(1,3)], dat[,c(1,4)], dat[,c(2,3)], dat[,c(2,4)], dat[,c(3,4)]. Can someone help me?

CodePudding user response：

I can't verify this approach since I don't know the rinform package. You could try

apply(combn(1:4, 2), 2, function(i) mutual_info(dat[, i], locale = TRUE))

combn(1:4, 2) creates a matrix of all possible of two elements from 1:4.
We use this as an input for the apply function to generate the result.

Naming

If you save the new data into a matrix

new_data <- apply(combn(1:4, 2), 2, function(i) mutual_info(dat[, i], locale = TRUE))

you could name the columns using the same approach:

colnames(new_data) <- apply(combn(1:4, 2), 2, function(i) paste0("V", i, collapse = ""))