for loop on a list of matrix and the mean of all the matrix in R

Question

I have a list that contain matrix named from 1 to 10 each one represent a year, so i have to generate matrix MG between M1 and M2, also between M2 et M3, M3 et M4....

so for example to calculate matrix betweem M1 and M2 i have to do this following calculation:

L=c(rowSums(Matrix1)) #(sum of each matrix'row)
K=c(rowSums(MatriX2))

G=L+K 

SM=Matrix1+Matrix2  #( sum of the two matrix)

MG=sweep(SM,1,G,FUN = "/") #( div SM by G )

output=list(MG)

and at the end generate a matrix that calculate the mean of all the MG included in the list.

I'm still new in R, any help would be appreciated thanks

Please provide a reproducible example that will go with your code. — Roman Luštrik, Feb 05 '18 at 12:13
The `c()` around `rowSums` is unnecessary, `rowSums` already returns a vector. — LAP, Feb 05 '18 at 12:20

score 0 · Answer 1 · answered Feb 05 '18 at 12:19

It's good that you have your several matrices into a list, we'll use Lapply on those to create a list of i-1 MG matrices, using your code (I didn't check it, just copied it into the function).

your_list <- list(M1 = as.matrix(iris[1:5,1:4]),
                  M2 = as.matrix(iris[6:10,1:4]),
                  M3 = as.matrix(iris[11:15,1:4]))  

your_function <- function(Matrix1,Matrix2){
  L=c(rowSums(Matrix1)) #(sum of each matrix'row)
  K=c(rowSums(Matrix2))
  G=L+K 
  SM=Matrix1+Matrix2  #( sum of the two matrix)
  MG=sweep(SM,1,G,FUN = "/") #( div SM by G )
}

MG_list <- lapply(1:(length(your_list)-1),function(i) your_function(your_list[[i]],your_list[[i+1]]))

Then to do an average of all these matrices we start by summing them, then divide by the number of matrices, see ?Reduce to understand how it works :

avg_MG <- Reduce(`+`,MG_list) / length(MG_list)

I have a question please, let's say x+y=0 and rowSums(x)+rowSums(y)=0 that will generate 0/0 which become NA, so the mean' calculation will be affected. do you have any idea how to replace all the NA by 0 or any other suggestion? thank you — S.ben, Feb 05 '18 at 21:07

score 0 · Accepted Answer · answered Feb 05 '18 at 12:41

An alternative to @Moody_Mudskippers approach would be to use Map.

Example data:

set.seed(1)
matlist <- list(matrix(sample(1:10, 4, replace = T), nrow = 2, ncol = 2),
                matrix(sample(1:10, 4, replace = T), nrow = 2, ncol = 2),
                matrix(sample(1:10, 4, replace = T), nrow = 2, ncol = 2),
                matrix(sample(1:10, 4, replace = T), nrow = 2, ncol = 2),
                matrix(sample(1:10, 4, replace = T), nrow = 2, ncol = 2),
                matrix(sample(1:10, 4, replace = T), nrow = 2, ncol = 2),
                matrix(sample(1:10, 4, replace = T), nrow = 2, ncol = 2),
                matrix(sample(1:10, 4, replace = T), nrow = 2, ncol = 2),
                matrix(sample(1:10, 4, replace = T), nrow = 2, ncol = 2),
                matrix(sample(1:10, 4, replace = T), nrow = 2, ncol = 2))
names(matlist) <- paste0("M", 1:10)

Generate output of consecutively swept matrices:

output <- Map(function(x, y){
  sweep(x+y, 1, rowSums(x)+rowSums(y), FUN = "/")
}, matlist[-10], matlist[-1])

Calculate means:

Reduce(`+`, output) / length(output)

          [,1]      [,2]
[1,] 0.4984006 0.5015994
[2,] 0.4730748 0.5269252

I have a question please, let's say x+y=0 and rowSums(x)+rowSums(y)=0 that will generate 0/0 which become NA, so the mean' calculation will be affected. do you have any idea how to replace all the NA by 0 or any other suggestion? thank you — S.ben, Feb 05 '18 at 21:06
You could just run `output <- lapply(output, function(x) ifelse(is.na(x), 0, x))` after generating the `output` and before calculating the means. — LAP, Feb 06 '18 at 07:58

for loop on a list of matrix and the mean of all the matrix in R

2 Answers2

Linked