I have the following dataset in R and I want to calculate Shannon's entropy. In order to do that since the data are continuous, I have to discretise them. Using the discretize2d function of Entropy package, the entropy between $X_1$ and $X_2$ can be calculated as follows:

set.seed(1234)
data <- matrix(rnorm(150 * 11, mean = 0, sd = 1), 150, 11)

library(entropy)
dis <- discretize2d(data[,1],data[,2], numBins1 = 10, numBins2 = 10)

entropy(dis)

I want to create a list containing all the discretize2d results between between the variables of data so i can later just use entropy(dis$1.2) and getting the same result as entropy(dis). Can someone help me code it?

CodePudding user response：

Here is an all-base R solution. We use the combn(x, m) function to generate all combinations of the elements of x with size m. Here we want pairs so m = 2. This creates a 2 by 55 matrix. Then use apply() to iteratively apply discretize2d() over columns of that matrix. The second argument of apply() is 2, meaning to apply over columns. We also specify simplify = FALSE so that the result will stay a list rather than being coerced to an array.

combs <- combn(1:ncol(data), 2)
dis <- apply(combs, 2, function(x) discretize2d(dat[, x[1]], dat[, x[2]], numBins1 = 10, numBins2 = 10), simplify = FALSE)

If you want names for the elements like you specified, such as dis$1.2, you can do this:

names(dis) <- apply(combs, 2, paste, collapse = '.')

Finally you could also calculate entropy for all elements at once with lapply():

lapply(dis, entropy)