Selecting odd and even columns in a matrix

General
1

Hi,
I would like to select odd and even columns from a matrix that in columns' names has got odd and even numbers as a strings.
Here is my_matrix:

my_matrix <- structure(c(NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, 
NA, NA, NA, NA, NA, NA, NA), .Dim = c(6L, 14L), .Dimnames = list(
    c("P18", "P19", "P20", "P21", "P22", "P23"), c("P35_cor", 
    "P35_p", "P36_cor", "P36_p", "P37_cor", "P37_p", "P38_cor", 
    "P38_p", "P39_cor", "P39_p", "P40_cor", "P40_p", "P41_cor", 
    "P41_p")))

I have tried with parse_number but it converts all variables to V1, V2 and so on.

colnames(my_matrix) <- names(parse_number(colnames(my_matrix)))

How do I do it ? Thank you for any ideas.

2 
my_matrix <- structure(c(
  NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,
  NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,
  NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,
  NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,
  NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, NA,
  NA, NA, NA, NA, NA, NA, NA
), .Dim = c(6L, 14L), .Dimnames = list(
  c("P18", "P19", "P20", "P21", "P22", "P23"), c(
    "P35_cor",
    "P35_p", "P36_cor", "P36_p", "P37_cor", "P37_p", "P38_cor",
    "P38_p", "P39_cor", "P39_p", "P40_cor", "P40_p", "P41_cor",
    "P41_p"
  )
))

# odd and even columns

# get indices of columns
get_index <- function(x,y) seq(from = y, to = dim(x)[2],by = 2)

my_matrix[,get_index(my_matrix,1)]
#>     P35_cor P36_cor P37_cor P38_cor P39_cor P40_cor P41_cor
#> P18      NA      NA      NA      NA      NA      NA      NA
#> P19      NA      NA      NA      NA      NA      NA      NA
#> P20      NA      NA      NA      NA      NA      NA      NA
#> P21      NA      NA      NA      NA      NA      NA      NA
#> P22      NA      NA      NA      NA      NA      NA      NA
#> P23      NA      NA      NA      NA      NA      NA      NA
my_matrix[,get_index(my_matrix,2)]
#>     P35_p P36_p P37_p P38_p P39_p P40_p P41_p
#> P18    NA    NA    NA    NA    NA    NA    NA
#> P19    NA    NA    NA    NA    NA    NA    NA
#> P20    NA    NA    NA    NA    NA    NA    NA
#> P21    NA    NA    NA    NA    NA    NA    NA
#> P22    NA    NA    NA    NA    NA    NA    NA
#> P23    NA    NA    NA    NA    NA    NA    NA

# odd and even column names
shadow <- as.numeric(gsub("[Pp_cor]","",colnames(my_matrix)))

my_matrix[,which(shadow %% 2 != 0)]
#>     P35_cor P35_p P37_cor P37_p P39_cor P39_p P41_cor P41_p
#> P18      NA    NA      NA    NA      NA    NA      NA    NA
#> P19      NA    NA      NA    NA      NA    NA      NA    NA
#> P20      NA    NA      NA    NA      NA    NA      NA    NA
#> P21      NA    NA      NA    NA      NA    NA      NA    NA
#> P22      NA    NA      NA    NA      NA    NA      NA    NA
#> P23      NA    NA      NA    NA      NA    NA      NA    NA
my_matrix[,which(shadow %% 2 == 0)]
#>     P36_cor P36_p P38_cor P38_p P40_cor P40_p
#> P18      NA    NA      NA    NA      NA    NA
#> P19      NA    NA      NA    NA      NA    NA
#> P20      NA    NA      NA    NA      NA    NA
#> P21      NA    NA      NA    NA      NA    NA
#> P22      NA    NA      NA    NA      NA    NA
#> P23      NA    NA      NA    NA      NA    NA
1 Like
3

This is great, works like a charm. Thank you.
If I may ask please:

  1. in the get_index function, what does it mean, I mean x are rows and y are columns ?
seq(from = y, to = dim(x)[2],by = 2)

and here:

"[Pp_cor]"

I think this is regex, why do you use Pp ?
Thank you again.

4

Yes, subsets work by row/column order, but x and y in the quoted code are the formal arguments to get_index() representing the name of the matrix, x and the positions from which to be the seq so that we end up with odd and even indices with which to work.

seq(from = 1, to = dim(mtcars)[2], by = 2) # the first one is passed as y
#> [1]  1  3  5  7  9 11
seq(from = 2, to = dim(mtcars)[2], by = 2) # the first one is passed as y
#> [1]  2  4  6  8 10

is called a character class regex that searches for the letters and characters ; here it is being used to remove those from the variable names as an argument to gsub.

5

Thank you, that's great.
So "p" stands for "35" in this regex character class ?
obraz

Why does it have to be "p" ?

6

p stands for "p", just one of the characters in the variable names that are non-numeric and are being removed to be able to specify those columns whose names, if their names contained only numbers could be identified as odd or even.

7

Thank you very much indeed, I just try to understand the magic behind this regex of yours [Pp_cor],
that it removes all letters and _ from both columns like "P35_cor" and "P35_p".

obraz

8

Hi @technocrat
I have tried a bit experimenting with your code and suddenly RStudio's behaviour became a bit strange.
When I write following line of code:

my_matrix[,which(shadow %% 2 != 0)] %>% apply(my_matrix[, odd], 2, function(x) {return(1)}) aaaaaa

at the end of a line cursor become a vertical line crossing letters "a".
obraz

obraz

It looks like Delete and Backspace keys are not deleting what I want as usual.
The cursor is standing right in the middle of "a" not before as usual.

Did I press something on my keyboard or is it a bug ?

9 
Error: unexpected symbol in "my_matrix[,which(shadow %% 2 != 0)] %>% apply(my_matrix[, odd], 2, function(x) {return(1)}) aaaaaa

should produce an error message

10

Yes, I know, I just have written it to show a bit strange behaviour of the cursor itself.

11

Here is a different solution. There is no requirement to use column names just because they are present.

odd <- seq(1, 14, 2)
my_matrix2<-matrix(0,nrow=6,ncol=7)
j=1
for (i in odd){
  my_matrix2[,j] <- my_matrix[ ,i]
  j=j+1
}
### now recover column names and reapply them
colnames(my_matrix2) <- colnames(my_matrix[,odd])
12

This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.

If you have a query related to it or one of the replies, start a new topic and refer back with a link.