R help - Oct 2006 - applying order() in a for loop

Hi,

with the code below I try to organize sections of a data.frame. The 
dataframe has 9 data columns. I'd like to order the columns of the 
data.frame by the sum of the columns' entries. The sections of the 
columns are identified by "i". The order of the columns is given by 
the vector "z". Each section of the data.frame is ordered 
differently. This can be seen when looking at the individual rows of 
"h" where all different "z" vectors are stored for later
use.

In a third for loop, "e" is calculated based on the ordering 
calculated before. "e" is the re-organized data.frame with now all 
column segments in descending order -- in theory.

Problem:
The code below does not update the "z" in the third for loop. It 
remains the initial "z" that I calculate the first time around. I 
tried re-initializing "z" at various places in the code, hoping to 
get rid of the old values but with no apparent consequences. Also in 
the help I could not find anything that I could relate to my problem.

Does anybody have a suggestion on how to get the appropriate "z" to 
work when shuffling the data around for "e"?

Thanks already for suggestions!

Kim

### Code ####

for (i in 11:19)
     {
         for (j in 3:ncol(data))
             {
                 a <- sum(data[,j][data[,1] == i/10])
                 b <- c(b,a)
             }

         z <- order(b, decreasing = TRUE)
         d <- length(data[,1][data[,1] == i/10])

         h <- rbind(h,z) ### Just to check whether I really get 
different orderings for the different data segments.
	
         for (l in 1:d)
             {
                 e <- data[((i-10)*d+l-d),(3:ncol(data))][z] ## This 
z does not get updated but remains the initially calculated z
                 f <- rbind(f,e)
             }
         b <- NULL
         e <- NULL
         z <- NULL
     }

Hi

your code is strange and not reproducible. 
We do not have data
h is not defined but you use it in h <- rbind(h,z)

Based on what you say about your problem, I would probably try to 
split data frame into list accorging to "i". Then I would shufle 
columns in each part of list and then I would use lapply and do.call 
to concatenate all together


lll <- split(DF, i)
sss <- lapply(lll, function(x) order(colSums(x)))

# here could come for cycle 
xxx<-vector("list",3)
for (i in 1:3) xxx[i]<-list(as.matrix(lll[[i]][,sss[[i]]]))
do.call("rbind",xxx)

HTH
Petr




    

On 26 Oct 2006 at 15:40, Kim Milferstedt wrote:

Date sent:      	Thu, 26 Oct 2006 15:40:33 -0500
To:             	r-help at stat.math.ethz.ch
From:           	Kim Milferstedt <milferst at uiuc.edu>
Subject:        	[R] applying order() in a for loop
> Hi,
> 
> with the code below I try to organize sections of a data.frame. The
> dataframe has 9 data columns. I'd like to order the columns of the
> data.frame by the sum of the columns' entries. The sections of the
> columns are identified by "i". The order of the columns is given
by
> the vector "z". Each section of the data.frame is ordered
differently.
> This can be seen when looking at the individual rows of "h" where
all
> different "z" vectors are stored for later use.
> 
> In a third for loop, "e" is calculated based on the ordering 
> calculated before. "e" is the re-organized data.frame with now
all
> column segments in descending order -- in theory.
> 
> Problem:
> The code below does not update the "z" in the third for loop. It
> remains the initial "z" that I calculate the first time around. I
> tried re-initializing "z" at various places in the code, hoping
to get
> rid of the old values but with no apparent consequences. Also in the
> help I could not find anything that I could relate to my problem.
> 
> Does anybody have a suggestion on how to get the appropriate "z"
to
> work when shuffling the data around for "e"?
> 
> Thanks already for suggestions!
> 
> Kim
> 
> ### Code ####
> 
> for (i in 11:19)
>      {
>          for (j in 3:ncol(data))
>              {
>                  a <- sum(data[,j][data[,1] == i/10])
>                  b <- c(b,a)
>              }
> 
>          z <- order(b, decreasing = TRUE)
>          d <- length(data[,1][data[,1] == i/10])
> 
>          h <- rbind(h,z) ### Just to check whether I really get 
> different orderings for the different data segments.
> 
>          for (l in 1:d)
>              {
>                  e <- data[((i-10)*d+l-d),(3:ncol(data))][z] ## This z
> does not get updated but remains the initially calculated z
>                  f <- rbind(f,e)
>              }
>          b <- NULL
>          e <- NULL
>          z <- NULL
>      }
> 
> ______________________________________________
> 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 and provide commented,
> minimal, self-contained, reproducible code.
Petr Pikal
petr.pikal at precheza.cz