[R] elements in each row of a matrix to the left.

[john.gavin@ubs.com]

Hi Patrick/Jeff,

> Does
> 
> t(apply(z, 1, sort, na.last=TRUE))
> 
> do what you want?

Not quite.

t(apply(z, 1, sort, na.last=TRUE))

     [,1] [,2] [,3]
[1,]    1    1   NA
[2,]    1    1   NA
[3,]    1    1   NA
[4,]    1   NA   NA
[5,]    1   NA   NA
[6,]   NA   NA   NA

Row 2 is the problem.

I dont want to move all NAs to the end of each row.
I just want to move all of the NAs before the first non-NA element,
if any, to the end of each row.
So in my example, rows 1 and 2 should remain unchanged.

What I have got at the moment is ugly

shiftLeft <- function(z)
{ x <- as.data.frame(t(z)) # work with cols not rows.
  ans <- lapply(x, function(xx) 
  { # get indices of first and last non-NA element
    ind <- which(!is.na(xx))
    ind <- ind[c(1, length(ind))]
    # if all NAs or if first element is non-NA do no work
    if (any(is.na(ind)) || ind[1] == 1) xx else
    { temp <- numeric(length(xx)) ; temp[] <- NA
      # move elements in posns ind[1] to ind[2] to the start
      temp[1:(ind[2]-ind[1]+1)] <- xx[ind[1]:ind[2]]
      temp
    } # if
  }) # lapply
  ans <- as.matrix(data.frame(ans))
  dimnames(ans) <- dimnames(z)
  t(ans)
}

> z ; shiftLeft(z)
     [,1] [,2] [,3]
[1,]    1    1   NA
[2,]    1   NA    1
[3,]   NA    1    1
[4,]   NA   NA    1
[5,]   NA    1   NA
[6,]   NA   NA   NA
     [,1] [,2] [,3]
[1,]    1    1   NA
[2,]    1   NA    1
[3,]    1    1   NA
[4,]    1   NA   NA
[5,]    1   NA   NA
[6,]   NA   NA   NA

I feel that there is probably a shorter vectorised way to do this.
In general, I have matrices (z) with several thousand rows and 
and few hundred columns so vectorisation would help.

Regards,

John.


> -----Original Message-----
> From: Patrick Burns [mailto:pburns at pburns.seanet.com]
> Sent: 27 February 2006 19:55
> To: Gavin, John
> Cc: r-help at stat.math.ethz.ch
> Subject: Re: [R] elements in each row of a matrix to the left.
> 
> John,
> 
> Does
> 
> t(apply(z, 1, sort, na.last=TRUE))
> 
> do what you want?
> 
> 
> Patrick Burns
> patrick at burns-stat.com
> +44 (0)20 8525 0696
> http://www.burns-stat.com
> (home of S Poetry and "A Guide for the Unwilling S User")
> 
> john.gavin at ubs.com wrote:
> 
> >Hi,
> >
> >Given a matrix like
> >
> >(z <- matrix(c(
> >1, 1, NA, NA, NA, NA,
> >1,  NA, 1,  NA, 1, NA,
> >NA, 1, 1,  1,  NA, NA), ncol = 3))
> >
> >     [,1] [,2] [,3]
> >[1,]    1    1   NA
> >[2,]    1   NA    1
> >[3,]   NA    1    1
> >[4,]   NA   NA    1
> >[5,]   NA    1   NA
> >[6,]   NA   NA   NA
> >
> >is there a vectorised way to produce the output like
> >
> >     [,1] [,2] [,3]
> >[1,]    1    1   NA
> >[2,]    1   NA    1
> >[3,]    1    1   NA
> >[4,]    1   NA   NA
> >[5,]    1   NA   NA
> >[6,]   NA   NA   NA
> >
> >That is, given an n by m matrix, and going row by row, 
> >if the first non-NA element is in column k
> >I want to move elements in columns from k to m
> >to columns 1 to m-k+1 with NAs filling in from 
> >m-k+2 to m.
> >
> >  
> >
> >>version
> >>    
> >>
> >         _              
> >platform i386-pc-mingw32
> >arch     i386           
> >os       mingw32        
> >system   i386, mingw32  
> >status                  
> >major    2              
> >minor    2.1            
> >year     2005           
> >month    12             
> >day      20             
> >svn rev  36812          
> >language R        
> >
> >Regards,
> >
> >John.
> >
> >John Gavin <john.gavin at ubs.com>,
> >Quantitative Risk Control,
> >UBS Investment Bank, 6th floor, 
> >100 Liverpool St., London EC2M 2RH, UK.
> >Phone +44 (0) 207 567 4289
> >Fax   +44 (0) 207 568 5352
> >
> >Visit our website at http://www.ubs.com
> >
> >This message contains confidential information and is 
> intend...{{dropped}}
> >
> >______________________________________________
> >R-help at stat.math.ethz.ch mailing list
> >https://stat.ethz.ch/mailman/listinfo/r-help
> >PLEASE do read the posting guide! 
http://www.R-project.org/posting-guide.html
>
>
>
>  
>


Visit our website at http://www.ubs.com

This message contains confidential information and is intend...{{dropped}}