Using Aggregate and Loop together

[Leonardo Baltazar]

My friends, I'm with a problem to put aggregate into a loop, I need execute the aggregate in differents situations, for example, I gonna do too much select's if, and in each one I need run a aggregate, but aggregate cannot run in a LOOP, exist another way to do this?

DEFINE Gini (!POSITIONAL !TOKENS(1))

SORT CASES BY !1 (A).

VECTOR Tri = Tri1 TO Tri199.
LOOP #I=1 TO 199.
SELECT IF Tri(#I) =1.

AGGREGATE OUTFILE = *
/PRESORTED
/BREAK = !1
/persons = N .
WEIGHT BY persons.

COMPUTE brk = 1.
AGGREGATE OUTFILE = 'C:\Rede_estagiário2\Gini\RENDAagg.sav'
/BREAK = brk
/sumRENDA = SUM(!1).

END LOOP.
EXECUTE.


!ENDDEFINE.

[Andy W]

You could do functionally what you say you want with looping constructs within the macro (i.e. !DO !varname=start !TO finish). Given your incomplete code snippet I suspect there is an easier way (e.g. your multiple aggregates look to me like they could be amenable to one aggregate with a sum - possibly very easily done with MATRIX over the entire 199 Tri variables).

Maybe if you step back a bit and describe what you are trying to do the group can give better advice for a workflow.

[David Marso]

Yes!!! Rather than expect others to sort and 'fix' through an undocumented mess which doesn't work, describe exactly what you are trying to do.  As Andy stated, there is the !DO !DOEND in Macro.
MATRIX is another possibility depending upon the data size etc.
I would just roll the data out using VARSTOCASES and use a SINGLE aggregate, or is that too simple?
OTOH:  You really don't provide any clue as to why you are doing this!
CONTEXT pal! CONTEXT!!!

Andy W wrote

You could do functionally what you say you want with looping constructs within the macro (i.e. !DO !varname=start !TO finish). Given your incomplete code snippet I suspect there is an easier way (e.g. your multiple aggregates look to me like they could be amenable to one aggregate with a sum - possibly very easily done with MATRIX over the entire 199 Tri variables).

Maybe if you step back a bit and describe what you are trying to do the group can give better advice for a workflow.

[Art Kendall]

Are you trying to get a gini coefficient for each of 199 variables!?!
On 9/13/2013 8:43 AM, Leonardo Baltazar [via SPSSX Discussion] wrote:

My friends, I'm with a problem to put aggregate into a loop, I need execute the aggregate in differents situations, for example, I gonna do too much select's if, and in each one I need run a aggregate, but aggregate cannot run in a LOOP, exist another way to do this?

DEFINE Gini (!POSITIONAL !TOKENS(1))

SORT CASES BY !1 (A).

VECTOR Tri = Tri1 TO Tri199.
LOOP #I=1 TO 199.
SELECT IF Tri(#I) =1.

AGGREGATE OUTFILE = *
/PRESORTED
/BREAK = !1
/persons = N .
WEIGHT BY persons.

COMPUTE brk = 1.
AGGREGATE OUTFILE = 'C:\Rede_estagiário2\Gini\RENDAagg.sav'
/BREAK = brk
/sumRENDA = SUM(!1).

END LOOP.
EXECUTE.


!ENDDEFINE.

---

If you reply to this email, your message will be added to the discussion below:

http://spssx-discussion.1045642.n5.nabble.com/Using-Aggregate-and-Loop-together-tp5721960.html

To start a new topic under SPSSX Discussion, email [hidden email]
To unsubscribe from SPSSX Discussion, click here.
NAML

[Richard Ristow]

On 9/13/2013 8:43 AM, Leonardo Baltazar [via SPSSX Discussion] wrote:

>I'm with a problem to put aggregate into a loop,
>I need execute the aggregate in differents
>situations, for example, I gonna do too much
>select's if, and in each one I need run a
>aggregate, but aggregate cannot run in a LOOP, exist another way to do this?

You want to run with 199 different BREAK
variables. Much the easiest way to do this (do I
surprise anyone, here?) is to use CASESTOVARS to
get *one* variable in 199 variable groups. Like this (but not tested):

VARSTOCASES  /MAKE V FROM V1 TO V199
  /INDEX = Var#(199)
  /KEEP =  ID
  /NULL = KEEP.

AGGREGATE OUTFILE = *
    /BREAK = INDEX V
    /persons = N .

WEIGHT BY persons.

AGGREGATE OUTFILE = 'C:\Rede_estagiário2\Gini\RENDAagg.sav'
    /BREAK = INDEX
    /sumRENDA = SUM(V).

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Richard Ristow]

Sorry - CORRECTION to the code I posted: use Var#
instead of INDEX, in code that follows
VARSTOCASES. (And still not tested, so there may be other problems.)

VARSTOCASES  /MAKE V FROM V1 TO V199
  /INDEX = Var#(199)
  /KEEP =  ID
  /NULL = KEEP.

AGGREGATE OUTFILE = *
    /BREAK = Var# V
    /persons = N .

WEIGHT BY persons.

AGGREGATE OUTFILE = 'C:\Rede_estagiário2\Gini\RENDAagg.sav'
    /BREAK = Var#
    /sumRENDA = SUM(V).

=====================
To manage your subscription to SPSSX-L, send a message to
[hidden email] (not to SPSSX-L), with no body text except the
command. To leave the list, send the command
SIGNOFF SPSSX-L
For a list of commands to manage subscriptions, send the command
INFO REFCARD

[Leonardo Baltazar]

Kendall,

The 199 variables are quarter's where I need calculate the Gini coefficiente for each, and to do this I need use agreggate in each quarter, but I didn't get do this with loop command, because the command has no support to aggregate, and I'm looking for another way.

[David Marso]

Why don't YOU read through your thread and heed the advice several have proposed?
VARSTOCASES blah blah blah...
Or do you insist on following your own dead end path with LOOP and AGGREGATE?

Leonardo Baltazar wrote

Kendall,

The 199 variables are quarter's where I need calculate the Gini coefficiente for each, and to do this I need use agreggate in each quarter, but I didn't get do this with loop command, because the command has no support to aggregate, and I'm looking for another way.