Re: st: Re: Analysis of text

[Taavi Lai <taavi.lai@ut.ee>]

Dear Michael,

Thank you for the suggestions and questions.

The problem is that I don't have much more detail at the moment as the 
files aren't handed over to me yet and the background documentation is 
almost missing. So I posted my original question to get first 
impressions whether I have to turn to a programmer for a specialised 
program or could I try to resolve it on my own.

Said that I can still answer some of your questions. The file structure 
is described to me as "free text, like an essay". The files are computer 
generated so the structure should be uniform. There are multiple lines, 
at least one per each question. The number of questions per person 
varies between 300 and 500 (overall number of variables ~700). Missing 
variables are left empty, but I don't know if the question numbers are 
shown or the "empty" means that those are omitted as well.

Best regards,
Taavi

Michael Blasnik wrote:

> Yes, it can be done.  You should really post more details if you want 
> more help.  How are the files structured?  Are there multiple lines in 
> each?  How many variables?  How consistent is the layout across 
> files?  How are missing values coded?  etc..  Why don't you just show 
> a section of a file?
>
> Without more info, I'd guess that you will most likely use the -file- 
> command to read in the data (but infix may work, perhaps with 
> -split-). Looping across files and accumulating results is fairly easy.
>
> M Blasnik
>
>
> ----- Original Message ----- From: "Taavi Lai" <taavi.lai@ut.ee>
> To: <statalist@hsphsun2.harvard.edu>
> Sent: Friday, December 08, 2006 1:56 PM
> Subject: st: Analysis of text
>
>
>
> > Dear statalisters,
> >
> > I have a set (~10 000) of text files, each containing questionnaire 
> > answers for one person. These files are structured as plain text 
> > files without any tabular form.
> > I�d like to loop through all these files and generate one datafile as 
> > a result. Say, "M1" represents a question number and the value for 
> > this question is thus the text/number between  "M1" and "M2".
> >
> > Could such a thing be done using Stata? Any suggestions and comments 
> > are welcome.
> >
> > Best regards,
> > Taavi
> *
> *   For searches and help try:
> *   http://www.stata.com/support/faqs/res/findit.html
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/