using vectors and matrixes

[Bogdan Voicu-2]

Dear all,

 

I have a repetitive task, which requires to compute several weighted means for each case in a data base.

More exactly, for each case there are 9 variables (zR011HF to zr019HF), that I need to sum up weighted with different weights, depending on the values (ranging from 1 to 6) of another variable (idd6bis).

 

The operation repeats with several other sets of variable and I expect that I will need to change several times the weights, when simulating the final model.

 

My idea was to build a series of commands like this one:

 

vector prim=zR011HF to zr019HF.

vector weights=

65           5              5              10           6              5              4              0              0

40           15           5              15           11           9              5              0              0

50           7              5              15           10           8              5              0              0

30           10           10           20           15           10           5              0              0

10           20           10           20           30           5              5              0              0

5              5              5              15           10           5              5              25           25.

loop #i=1 to 6.

  loop #j=1 to 9.

                compute k=#j+(#i-1)*#j.

               if idd6bis=#i zR01HF=sum(zR01HF, weights(k)*prim(#j)).

  end loop.

end loop.

exe.

del var k.

 

The above solution does not work. I cannot use the vector command for the weights. My question is which command should I use to declare the weights as a matrix/vector, using the syntax file.

 

Thank you in advance!

Bogdan

 

 

 

---

Bogdan Voicu

Principal research fellow, Romanian Academy of Science, Research Institute for Quality of Life

Associate professor, Department of Sociology, Lucian Blaga University of Sibiu

[hidden email]; http://www.bogdanvoicu.ro

 

 

 

[Rich Ulrich]

Maybe the Matrix solution can work, though I don't see how it takes into account
the test in the example on idd6bis.

I could be uninformed, but I don't know a compact way to initialize the values of Vector
in SPSS except to a single value ("0", usually), where Recode works.   I think it takes
separate Computes.  So here are some comments, first, on how to get around that
limitation.

For your problem, where you also want to have a program that you can read, I think
I would have a Compute which sets up a text-array with the weights, using 54F3 for
the systematic spacing.  Then, I would use the internal, formatted Read (Decode?) 
to create 54 variables which are read into the vector of #Weight.  -- If you start the
name with "#",  then the values are preserved from record to record, so you can
use a "#-name", and make the program more efficient by using DO IF $CASE=1
so that the reading only gets done once.

That is how I would import the values.

There are a couple of other weaknesses with the proposed code.
 - The outer loop on #i  is not needed - Instead of using the Loop/ IF  combination,
you will get *exactly*  the same result by using
+COMPUTE #i = idd6bis     ; to set the value of #i.

- You can use the SUM( ) function as you do, but I would initialize to make *sure*
that zR01HF is zero.  And that the code shows it.

What you want, after

vector prim=zR011HF to zr019HF.         is

RECODE zR011HF to zr019HF(else=0);  to initialize.

--
Rich Ulrich

---

Date: Wed, 25 May 2011 16:36:03 +0300
From: [hidden email]
Subject: using vectors and matrixes
To: [hidden email]

Dear all,

 

I have a repetitive task, which requires to compute several weighted means for each case in a data base.

More exactly, for each case there are 9 variables (zR011HF to zr019HF), that I need to sum up weighted with different weights, depending on the values (ranging from 1 to 6) of another variable (idd6bis).

 

The operation repeats with several other sets of variable and I expect that I will need to change several times the weights, when simulating the final model.

 

My idea was to build a series of commands like this one:

 

vector prim=zR011HF to zr019HF.

vector weights=

65           5              5              10           6              5              4              0              0

40           15           5              15           11           9              5              0              0

50           7              5              15           10           8              5              0              0

30           10           10           20           15           10           5              0              0

10           20           10           20           30           5              5              0              0

5              5              5              15           10           5              5              25           25.

loop #i=1 to 6.

  loop #j=1 to 9.

                compute k=#j+(#i-1)*#j.

               if idd6bis=#i zR01HF=sum(zR01HF, weights(k)*prim(#j)).

  end loop.

end loop.

exe.

del var k.

 

The above solution does not work. I cannot use the vector command for the weights. My question is which command should I use to declare the weights as a matrix/vector, using the syntax file.

 

[David Marso]

"I could be uninformed, but I don't know a compact way to initialize the values of Vector
in SPSS except to a single value ("0", usually), where Recode works."

Something like the following untested...
VECTOR #V(54).
DO IF $CASENUM=1.
DO REPEAT Index=1 TO 54
   / WTS=65           5              5              10           6              5              4              0              0    40           15           5              15           11           9              5              0              0    50           7              5              15           10           8              5              0              0    30           10           10           20           15           10           5              0              0    10           20           10           20           30           5              5              0              0     5              5              5              15           10           5              5              25           25.
COMPUTE #V(Index)=WTS .
END REPEAT .
END IF.
..........