Parsing in PASW

[Mark Webb-5]

I have received multi-mention market research data in the following format -
String        1 2 3 4 5  all in one variable.
I want to have numeric with each number in a separate variable.
The number of mentions can vary from respondent to respondent.
eg          Resp 1     1 2 3 4 5            Five mentions
              Resp 2     3 4                     two mentions

Is there any syntax/advice out there that someone is prepared to share ?
Regards

--
Mark Webb

[phone]
[phone]
Skype - webbmark
[hidden email]

=====================
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

[Albert-Jan Roskam]

Hi,

Is your source file a fixed format ascii like file? You could try to read it like that using the GUI, and generate a syntax that looks something like:
GET DATA  /TYPE = TXT
 /FILE = 'C:\temp\somefile.txt'
 /FIXCASE = 1
 /ARRANGEMENT = FIXED
 /FIRSTCASE = 2
 /IMPORTCASE = ALL
 /VARIABLES =
 /1  V5 0-0 A1
 V2 1-7 A7
 V3 8-11 A4
 V4 12-51 A40
 .
CACHE.
EXECUTE.

If that's not an option, perhaps you can use substrings:
compute var1 = number(substr(sourcevar, 1, 1), f1).
etc., up to var5

Cheers!!
Albert-Jan


--- On Fri, 10/2/09, Mark Webb <[hidden email]> wrote:

> From: Mark Webb <[hidden email]>
> Subject: [SPSSX-L] Parsing in PASW
> To: [hidden email]
> Date: Friday, October 2, 2009, 12:32 PM
> I have received multi-mention market
> research data in the following format -
> String        1 2 3 4 5  all in
> one variable.
> I want to have numeric with each number in a separate
> variable.
> The number of mentions can vary from respondent to
> respondent.
> eg          Resp 1
>    1 2 3 4 5
>     Five mentions
>              Resp
> 2     3 4
>              two
> mentions
>
> Is there any syntax/advice out there that someone is
> prepared to share ?
> Regards
>
> --
> Mark Webb
>
> [phone]
> [phone]
> Skype - webbmark
> [hidden email]
>
> =====================
> 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
>

=====================
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

[Brian Moore-7]

Hi Mark-
I've dealt with things like this before too & took the leap that you'll
need at some point both the # of mentions form each respondent & counts
for which.
The following does both.  It is somewhat flexible [put all your other
identifiers between "first" and "last"], but might make sense to do a
loop / do repeat type solution if you have a much larger set of
potential choices than the sample data.

Regards,
Brian

Simply paste all of what is below into a new instance of SPSS in the
syntax window to test *****KEEP YOUR ORIGINAL DATA SAFE*****

Data list list / thestringvar (a5) . /* read in some sample data .
Begin data
123
124
125
1
3
42
15
End data.


***file note - you didn't say if the strings are always concatenations
in numerical order, but this solution does not require that

compute flag=0.
compute #flagFIRST=INDEX(thestringvar,'always empty - just to allow
first to last processing')>0.
if #flagFIRST=1 flagFIRST=1.
form flagFIRST (f1.0).


compute flag=0.
compute #flagChoice1=INDEX(thestringvar,'1')>0.
if #flagChoice1=1 flagChoice1=1.
form flagChoice1 (f1.0).


compute flag=0.
compute #flagChoice2=INDEX(thestringvar,'2')>0.
if #flagChoice2=1 flagChoice2=1.
form flagChoice2 (f1.0).

compute flag=0.
compute #flagChoice3=INDEX(thestringvar,'3')>0.
if #flagChoice3=1 flagChoice3=1.
form flagChoice3 (f1.0).


compute flag=0.
compute #flagChoice4=INDEX(thestringvar,'4')>0.
if #flagChoice4=1 flagChoice4=1.
form flagChoice4 (f1.0).


compute flag=0.
compute #flagChoice5=INDEX(thestringvar,'5')>0.
if #flagChoice5=1 flagChoice5=1.
form flagChoice5 (f1.0).



compute flag=0.
compute #flagLAST=INDEX(thestringvar,'always empty - just to allow first
to last processing')>0.
if #flagLAST=1 flagLAST=1.
form flagLAST (f1.0).


Freq flagfirst to flaglast.
exe.

COMPUTE CountNMentions = sum.1(flagFIRST to flagLAST) .
EXECUTE .



Freq CountNMentions.
exe.




-----Original Message-----
From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of
Mark Webb
Sent: Friday, October 02, 2009 3:32 AM
To: [hidden email]
Subject: Parsing in PASW

I have received multi-mention market research data in the following
format -
String        1 2 3 4 5  all in one variable.
I want to have numeric with each number in a separate variable.
The number of mentions can vary from respondent to respondent.
eg          Resp 1     1 2 3 4 5            Five mentions
              Resp 2     3 4                     two mentions

Is there any syntax/advice out there that someone is prepared to share ?
Regards

--
Mark Webb

[phone]
[phone]
Skype - webbmark
[hidden email]

=====================
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

=====================
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