Question

My problem is the following:

I need to reduce a matrix cutting some columns away but keeping the names of column vectors.
DTM is my original matrix that looks like the following:

>DTM
   word1    word2    word3    word4
[1] 1         1        0        0
[2] 2         0        1        1
[3] 0         1        0        2

and I want to obtain a new matrix (DTMr in the following chunk of code) that has ‘labels’ and eliminates all columns whose sum of members is less than a threshold (say 2):

   word1    word4
[1] 1         0
[2] 2         1
[3] 0         2

>DTMr <- matrix(,nrow=nrow(DTM),ncol=d) # This should be the reduced matrix

where d is the number of columns of DTM that are larger than the threshold

>c = 1 # new counter
>for (col in 1:ncol(DTM))
>{
>  if (sum(DTM[,col]) > 2) 
>  { 
>    DTMr[,c] = DTM[,col]
>    
>    c=c+1
>  }
>}

Unfortunately in this way, DTMr is perfect, but it loses all labels (word 1, …word n).

Any ideas?

Claudio

Answer 1

A simple solution using subsetting and colSums:

Create some sample data:

set.seed(1)
x <- matrix(sample(0:2, 12, replace=TRUE), ncol=4)
colnames(x) <- LETTERS[1:4]
x
     A B C D
[1,] 0 2 2 0
[2,] 1 0 1 0
[3,] 1 2 1 0

Subset:

x[, colSums(x)<4]
     A D
[1,] 0 0
[2,] 1 0
[3,] 1 0