sorting a data.frame using a vector

From: Peter Dalgaard

Toralf Kirsten <tkirsten at izbi.uni-leipzig.de> writes:

Hi all,
I'm looking for an efficient solution (speed and memory) for the
following problem:
Given
- a data.frame x containing numbers of type double
   with nrow(x)>ncol(x) and unique row lables and
- a character vector y containing a sorted order labels

Now, I'd like to sort the rows of the data.frame x w.r.t. 
the order of
labels in y.

example:
x <- data.frame(c(1:4),c(5:8))
row.names(x)<-LETTERS[1:4]
y <- c("C","A","D","B")

My current solution is like this:
if(!is.null(y) && is.vector(y)) {
    nObj <- length(y)
    for (i in 1:nObj) {
      sObj <- y[i]
      k <- c(1:nrow(x))[row.names(x)==sObj]
      if (i != k) {
        names <- row.names(x)
        tObj <- row.names(x[i,])
        temp <- x[i,]
        x[i,] <- x[k,]
        x[k,] <- temp
        names[i] <- sObj
        names[k] <- tObj
        row.names(x) <- names
     }
   }
}

But I'm not happy with it because it is not really efficient. Any
other suggestions are welcome!
Anything wrong with x[y,] ???
Well... sometimes:
nm <- as.character(sample(1:1e5))
x <- data.frame(x1=rnorm(1e5), row.names=1:1e5)
system.time(x[nm, , drop=FALSE], gcFirst=TRUE)
[1] 155.13   0.01 156.10     NA     NA
system.time(x2<-x[match(nm, rownames(x)), , drop=FALSE], gcFirst=TRUE)
[1] 0.37 0.00 0.37   NA   NA
all(rownames(x2) == nm)
[1] TRUE
R.version
_              
platform i386-pc-mingw32
arch     i386           
os       mingw32        
system   i386, mingw32  
status                  
major    2              
minor    0.1            
year     2004           
month    11             
day      15             
language R              

Cheers,
Andy
-- 
   O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: 
(+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: 
(+45) 35327907

______________________________________________
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

"Liaw, Andy" <andy_liaw at merck.com> writes:
But I'm not happy with it because it is not really efficient. Any
other suggestions are welcome!
Anything wrong with x[y,] ???
Well... sometimes:

nm <- as.character(sample(1:1e5))
x <- data.frame(x1=rnorm(1e5), row.names=1:1e5)
system.time(x[nm, , drop=FALSE], gcFirst=TRUE)
[1] 155.13   0.01 156.10     NA     NA
system.time(x2<-x[match(nm, rownames(x)), , drop=FALSE], gcFirst=TRUE)
[1] 0.37 0.00 0.37   NA   NA
all(rownames(x2) == nm)
[1] TRUE
Yes, the internals are using  

   pmatch(i, rows, duplicates.ok = TRUE)

and pmatch() is a horrible lot slower than match(). Anyone for a spot
of hardcore optimization?

(Partial matching of character indices is a feature long regretted by
its inventors, but every time we consider killing it, we tend to recoil
in horror upon realizing what it would break...)
O__  ---- Peter Dalgaard             Blegdamsvej 3  
  c/ /'_ --- Dept. of Biostatistics     2200 Cph. N   
 (*) \(*) -- University of Copenhagen   Denmark      Ph: (+45) 35327918
~~~~~~~~~~ - (p.dalgaard at biostat.ku.dk)             FAX: (+45) 35327907