De-aggregation of data

General
1

Hi, I want to know which software can do de-aggregation of data ? For example I have got aggregated data and want to go to raw/individual data, in R I could use uncount() function, but is it possible to do it in other statistical software ?

obraz

obraz

2

I would expect this to be achievable in every programming language... perhaps you could narrow your question?

3

I would like to do it somehow in SPSS. Practically in every software there is aggregate function, but going back
to raw data is not possible. I work often with SPSS and I can import data (sav file) into R and uncount() it and then save it as sav file and be back in SPSS. The thing is that in SPSS factors can be numbered starting from 0 but in R indexing starts from 1 so I have to correct it in SPSS.
Is there a way maybe to start numbering levels from 0 in R or is it possible to uncount data in SPSS (using syntax) ?
I checked Jamovi and Jasp as well but did not find such an option there.

4

This is a little kludgey, but if the factor in R is named V you could do

V1<-as.numeric(V)-1

and then save V1 to SPSS.

5

It works but it removes factor's labels. What to do to retain labels ?

danne <- structure(list(condition = structure(c(1L, 1L, 1L, 1L, 2L, 2L, 
2L, 2L, 3L, 3L, 3L, 3L), levels = c("absent", "mild", "severe"
), class = "factor"), bloodtype = structure(c(1L, 2L, 3L, 4L, 
1L, 2L, 3L, 4L, 1L, 2L, 3L, 4L), levels = c("A", "B", "AB", "O"
), class = "factor"), count = c(543, 211, 90, 476, 44, 22, 8, 
31, 28, 9, 7, 31)), class = "data.frame", row.names = c(NA, -12L
), variable.labels = structure(character(0), names = character(0)), codepage = 1252L)

so basically I am asking, is this possible:

obraz
obraz518×689 56.2 KB

6

I feel you might get more help posting on an SPSS forum.

7

I havent used spss, so i asked LLM for example to uncount using loop.

I asked to start with a row of a representing one entry with an associated value 2, and a row for b representing 2 entries with a value 5.

DATA LIST FREE / Item (A1) Count Value.
BEGIN DATA
a 1 2
b 2 5
END DATA.

SORT CASES BY Count (D).
VECTOR newItem(100) / newValue(100).
COMPUTE idx = 1.
LOOP #i = 1 TO N.
  + DO REPEAT #j = 1 TO Count.
    + COMPUTE newItem(idx) = Item.
    + COMPUTE newValue(idx) = Value.
    + COMPUTE idx = idx + 1.
  + END REPEAT.
END LOOP.
EXECUTE.

VARSTOCASES /MAKE Item FROM newItem1 TO newItem100 /MAKE Value FROM newValue1 TO newValue100 /INDEX=idx.
SELECT IF NOT MISSING(Item).
EXECUTE.
8

Thank you,
how would it go for my data :
obraz

What should I copy/paste into that syntax of yours ?

9

Nobody knows the answer on that forum.