SUCCESS (Re: Noob needs help with simple script)

[David Spaugh]

Ben et al:

Success today, thanks!

I placed the 'set workspace' command as the 1st line of my script, but this 
did not resolve the problem.

I inserted an additional command into the syntax: IMPORTCASE=FIRST 5.

This successfully created a "spreadsheet" with 5 rows of data.  The Output 
Viewer contained a list of complaints for each row: three fields were 
non-numeric; improper numeric syntax, etc.  PSPP did not like a decimal that 
was in these otherwise numeric fields.

I increased the Importcase limit to 50 and then 500 and then 2000 records. 
At 2000, the computer required 20 seconds to succeed, and the stack of 
complaints on the Output page was obviously quite long.

I changed the F parameter to A for the three suspect fields, and expanded 
Imortcase to ALL.

I had my complete 28000 line sheet in less than 2 seconds.  Bob's your 
uncle.

I suspect that even with the Set Workspace command in the script, my Windows 
PC simply choked on the repeating "non-numeric" errors to the point of 
locking up.

The solution was reached by limiting the scope of the Get Data so that I 
could at least get some output to allow identification of the problem.

Okay then.  Thanks very much for your work, and your support!

While reading and learning today, I perused the Linux-suite offerings 
available through gnu.  I swear someday when I have time I will abandon 
Windows forever and join the unencumbered for a better computing experience.

DS

----- Original Message ----- 
From: "Ben Pfaff" <address@hidden>
To: "David Spaugh" <address@hidden>
Cc: <address@hidden>
Sent: Wednesday, May 11, 2011 9:11 PM
Subject: Re: Noob needs help with simple script


> "David Spaugh" <address@hidden> writes:
>
> > http://www.cdc.gov/nchs/ahcd/ahcd_questionnaires.htm
> >
> > Scroll down to Public-use Data Files / Downloadable Data Files.
> > Click the link for NAMCS, 1993-2008.
> > Choose NAMCS08.exe.
> > This downloads a 2.2MB zip file to your drive, which extracts to a 27MB 
> > data
> > file.  That's the file I'm working with.
>
> OK.  I did this and tried out the syntax that you provided.  It
> worked out OK for me, yielding a data window with 28741 cases.  I
> didn't see the problem that you mentioned.  (I'm running this on
> GNU/Linux; I think you said you were using Windows.)
>
> This suggests to me that the problem that you are seeing is one
> of the following:
>
>        * It is a bug in PSPP that has been fixed since 0.7.5.
>          You could try with the latest PSPP version to find this
>          out.
>
>        * It is a Windows-specific bug in PSPP that I am trying
>          to track down anyway.  More information at
>          https://savannah.gnu.org/bugs/?33250.  If this is the
>          problem, then running 'set workspace=100000000.' before
>          you load the data will make it go away.
>
>        * Something new...
>
> I'd be happy to try out your full syntax, if you like, too.
>
> > FYI, FWIW, etc:  If you surf the links for related documentation, you'll 
> > see
> > that gov actually provided an SPSS script for file import for a few 
> > data-sets
> > from 2006 and earlier.  That was actually my personal discovery point for
> > PSPP, leading to my attempt to create a simple script for the 2008 data 
> > file,
> > and to this email.  My goal is much simpler than the "segregate and 
> > label"
> > functions provided by the huge gov script.  I only need to import the 
> > data
> > into a usable PSPP "spreadsheet" to sort & determine the number of 382.x
> > diagnoses, and then what proportion of those were accompanied by a 
> > certain
> > prescription drug code.  I'll then adapt & apply the same script to 
> > multiple
> > prior years and to evaluate trends in antibiotic drug consumption for 
> > that
> > diagnosis.
>
> I tried running the .sps file for the 2006 data set here, to make
> sure that PSPP can handle it.  It mostly worked.  The only
> problems were either genuine syntax errors in the .sps file
> (e.g. trying to put a single-quoted inside a single-quoted string
> without doubling it) or using commands that are not yet
> implemented (e.g. ORTHOPLAN).
> --
> Ben Pfaff
> http://benpfaff.org