[R] Filling in empty arrays/lists from using "paste" function

[David Winsemius]

On Aug 25, 2009, at 3:52 AM, Uwe Ligges wrote:

> I highly suggest to rethink your problem in a way that you store the  
> things labelled, e.g., TA1 to TA5 (as well as all the others) as the  
> elements of a list TA. This way you have just one object TA that can  
> easily be accessed by index operations and the code looks much  
> cleaner in the end.
>
> Best,
> Uwe Ligges

Partial solutions considered?

TA <- by(a$pro, a$id, table)  # creates the desired format

But has 1's in the locations where the values of sal should be.

for (i in names(TA) ) {   # these are just the id's from "a"
    for (j in levels(a$pro) ) {
        TA[[i]][[j]] <- a[a$id==i & a$pro == j, "sal"] }}  #wrong

Doesn't work yet because I have not figured out the right combination  
of indexing on the r.h.s of the assignment. The logical test creates a  
vector that is not registered with the right values of sal somehow.  
Also the none-existent values get numeric(0) values. I'm thinking  
that  there may be a way for setting up a list with id and pro as  
indices and sal values as entries  and then reading from that  
temporary list.

ll <-ist()
 > for (i in 1:nrow(a) ) { ll[[ as.character(a$id[i]) ]] 
[[ as.character(a$pro[i]) ]] <- a$sal[i] }
 > ll
$idA1
bb gg
  0  1

$idA2
uu tt
  0  2

$idA3
ee bb
  4  4

$idA4
tt gg
  3  0

$idA5
uu ee
  0  1

But then running the nested for loop causes an error when I try to  
access non-existent ll subscripts. So I need an appropriate list  
traversal function that will "know" where it is on the list so it can  
use the subscript for an assignment from ll to TA. As yet have not  
cracked that nut.

>
>
> Steven Kang wrote:
>> Dear R users,
>> I am trying to fill in arrays (5 different according to distinct  
>> "id")
>> from objects produced from arbitrary data set below.
>>  a <-
>> data.frame(id=rep(c("idA1","idA2","idA3","idA4","idA5"), 
>> 2 
>> ),pro 
>> = 
>> c("bb","uu","ee","tt","uu","gg","tt","bb","gg","ee"),sal=rpois(10,2))
>>      id    pro sal
>> 1  idA1  bb   2
>> 2  idA2  uu   0
>> 3  idA3  ee   3
>> 4  idA4  tt   2
>> 5  idA5  uu   4
>> 6  idA1  gg   3
>> 7  idA2  tt   0
>> 8  idA3  bb   1
>> 9  idA4  gg   0
>> 10 idA5  ee   5
>> My desired outputs (5 arrays/lists classified according to distinct  
>> "id"
>> field) are as follow:
>>> TA1
>> bb ee gg tt uu
>> 2  0  3  0  0
>>> TA2
>> bb ee gg tt uu
>> 0  0  0  0  0
>>> TA3
>> bb ee gg tt uu
>> 1  3  0  0  0
>> ...... similarly for TA4 & TA5.


>> snipped

David Winsemius, MD
Heritage Laboratories
West Hartford, CT