Re: st: Want to load only part of ASCII file

[Jon Gettman <jgettman@hughes.net>]

When you open a delimited file in Excel you get a Text Import Wizard. 
The third stage of the wizard allows you to skip selected 
columns/variables when loading the file.  So one solution is to load 
the file in Excel, only select the variables you need, and then save 
the resulting file as a text(tab) delimited text file.  Then use the 
insheet command to load this into Stata.  (Make sure your vars in 
Excel are formatted appropriately as text or numbers, and the format 
will carry over.  If a numerical var shows up in State as a string 
variable, it's because it has a non-numerical character in a record.)

If you record this process as a Macro in Excel and examine the code 
you will discover that the programming is very much like the use of a 
Stata dictionary.  This suggests an alternate course of 
action.  Write a dictionary file only referencing the columns you 
wish to load, and then use the infile command to load the ASCII 
file.  The dictionary specifies the column location, name, format, 
and length of the data sought.  Infile is the real way to go here.

Jon Gettman






At 09:27 AM 7/21/2010, you wrote:
> Sometimes the best solution isn't a Stata solution; for this kind of
> problem I usually invoke some other software. For instance,
> Stat/Transfer will convert a .csv file to a Stata file while retaining
> only selected variables, and does not require vast RAM. Or, you can
> open in Excel, delete a few columns, and save as a new .csv file;
> Excel also buffers to disk and hence needs less RAM. (If you go the
> Excel route, though, be exceedingly careful, because Excel makes it
> too easy to corrupt your data  - for instance, by sorting a single
> column.)
>
> hth,
> Jeph
>
>
> On 7/20/2010 7:38 PM, enewton@notes.cc.sunysb.edu wrote:
> > Hi,
> >
> > I'm new to Stata, and new to this listserv. I'd like to load a very
> > large ASCII .csv file into a .dta Stata file but it keeps bumping up
> > against the 1.1g memory limit and Stata suggests I "drop some
> > observations or variables." My question is how one can load just a
> > portion of an ASCII file--for example, the specific variables of
> > interest, or only those observations where variable1=xyz--so as to
> > end up with a more manageable Stata file.
> >
> > A second question is why I can't set memory higher than about 1.3g,
> > when I have about 100g of space left on the computer, but that's
> > probably something my computer&  I need to work out between
> > ourselves.
> >
> > Thanks.
> >
> > * *   For searches and help try: *
> > http://www.stata.com/help.cgi?search *
> > http://www.stata.com/support/statalist/faq *
> > http://www.ats.ucla.edu/stat/stata/
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/