Align the data across variables

[jagadishpchary]

Hi:

I have data for 8 variables which i want to move the data across the variables using the syntax (filling the empty gaps). However I could find few methods like merging the data by using concatenation and then splitting into different variables - which really works but lot of coding involved. So does anyone can provide me a simple SPSS code which will work on the attached data "Before" and give the same results as shown "After" in the attached.

thanks.  

[Jignesh Sutar]

Something like this would work, if you isolate just those variable of interest and then merge back the results into the original dataset:

/******************************/.
data list free / V1 V2 V3 V4 V5 V6 V7 V8.
begin data.
20,10, ,3,13,21,250,
716,250, , , ,17,128,1
20,21,22, ,3,250, ,
end data.
dataset name dsSim.
compute caseid=$casenum.
/******************************/.

dataset copy dscopy.
dataset activate dscopy.
varstocases /make V from v1 to v8 /index=id(v).
casestovars /id=caseid /separator="".
add files file=* /keep=caseid v1 to v6.

[jagadishpchary]

Jignesh:

Thanks for providing the code.

We were able to get the results but can we retain the blank cases.

thanks.

[Jignesh Sutar]

Alternative solution below using VECTOR/LOOP without having to restructure dataset;

vector v=V1 to V8.
loop #i=1 to 8.
   loop #y=1 to nmiss(V1 to V8).
       do if (nmiss(V(#i))=1).
       loop #j=#i to 7.
           compute v(#j)=v(#j+1).
           if (#j+1=8) v(#j+1)=$sysmis.
       end loop.
       end if.
   end loop.
end loop.
exe.

[jagadishpchary]

Thanks Jignesh!!..this is the code i was looking for.

[Andy W]

Here is a looping approach I find alittle simpler to grok.


***************************************.
VECTOR X(8).
VECTOR V = V1 TO V8.
COMPUTE #C = 1.
LOOP #i = 1 TO 8.
  DO IF NOT MISSING(V(#i)).
    COMPUTE X(#C) = V(#i).
    COMPUTE #C = #C + 1.
  END IF.
END LOOP.
MATCH FILES FILE = * /DROP V1 TO V8.
RENAME VARIABLES (X1 TO X8 = V1 TO V8).
***************************************.

And here is a python solution as well. (Below just replace the RESULT part with new variables if you don't want to destroy the originals. For the above code just delete the last two lines.)

***************************************.
BEGIN PROGRAM Python.
def noneStrip(L):
  return [x for x in L if x is not None]
END PROGRAM.

SPSSINC TRANS RESULT = V1 TO V8
  /VARIABLES V1 TO V8
  /FORMULA "noneStrip([<>])".
***************************************.

[David Marso]

Similar to Andy's but in place rather than creating a new vector.

data list free / V1 V2 V3 V4 V5 V6 V7 V8.
begin data.
20,10, ,3,13,21,250,
716,250, , , ,17,128,1
20,21,22, ,3,250, ,
 , , , , , , , , ,
end data.


VECTOR V = V1 TO V8.
LOOP #i = 1 TO 7.
+  DO IF MISSING(V(#i)).
+    LOOP #j=#i + 1 TO 8.
+      DO IF NOT (MISSING(V(#j))).
+        COMPUTE V(#i)=V(#j).
+        COMPUTE V(#j)=$SYSMIS.
+      END IF.
+    END LOOP IF NOT(MISSING(V(#i))).
+  END IF.
END LOOP.
EXECUTE.

Andy W wrote

Here is a looping approach I find alittle simpler to grok.


***************************************.
VECTOR X(8).
VECTOR V = V1 TO V8.
COMPUTE #C = 1.
LOOP #i = 1 TO 8.
  DO IF NOT MISSING(V(#i)).
    COMPUTE X(#C) = V(#i).
    COMPUTE #C = #C + 1.
  END IF.
END LOOP.
MATCH FILES FILE = * /DROP V1 TO V8.
RENAME VARIABLES (X1 TO X8 = V1 TO V8).
***************************************.

And here is a python solution as well. (Below just replace the RESULT part with new variables if you don't want to destroy the originals. For the above code just delete the last two lines.)

***************************************.
BEGIN PROGRAM Python.
def noneStrip(L):
  return [x for x in L if x is not None]
END PROGRAM.

SPSSINC TRANS RESULT = V1 TO V8
  /VARIABLES V1 TO V8
  /FORMULA "noneStrip([<>])".
***************************************.