Convert 3-column long format to 1000 of upper (or lower) triangular matrices-CodePudding

I would like to convert three defined columns (data frame) in one thousand of upper (or lower) matrices of 8 traits each in R. Each matrix is written out row-wise in the dataframe. This is an example:

Dataframe with 3 columns (and 12000 rows). From this I need to have 1000 matrices (triangular):
           col1  col2  col3
row1         1     2     3 
row2         2     3     4
row3         7     4     5
row4         1     4     5
row5         2     4     5
row6         3     4     5
.            .     .     .
.            .     .     . 
.            .     .     . 
row12000     1     6     9

I need to convert them to 1000 matrices (triangular) of 8 traits each and put them in a list: Example of converting the data in the dataframe into the first matrix:


            trait 1     trait2         trait3    trait4 trait5 trait6  trait7 trait8
trait1: 1(row1 col1)  
trait2: 2(row1 col2)  5(row3 col3)  
trait3: 3(row1 col3)  1(row4 col1)  3(row6 col1)  
trait4: 2(row2 col1)  4(row4 col2)  4(row6 col2)  value  
trait5: 3(row2 col3)  5(row4 col3)  5(row6 col3)  value  value 
trait6: 4(row2 col3)  2(row5 col1)  .             value  value  value 
trait7: 7(row3 col1)  4(row5 col2)  .             value  value  value  value  
trait8: 4(row3 col2)  5(row5 col3)  .             value  value  value  value  value

CodePudding user response：

Collapse rowwise the matrix value the dataframe (df) value in a vector (all_values).
Create group of 36 values and split them in a list.
Place each of the 36 value in a lower triangular matrix of 8 X 8 and get a list of matrices.

all_values <- c(t(df))
dummmy_matrix <- matrix(nrow = 8, ncol = 8, 
                 dimnames = list(paste0('trait', 1:8), paste0('trait', 1:8)))

matrix_list <- lapply(split(all_values, ceiling(seq_along(all_values)/36)), function(x) {
  dummmy_matrix[lower.tri(dummmy_matrix, diag = TRUE)] <- x
  dummmy_matrix
})
matrix_list