Re: st: Dropping Alphanumeric elements from variables

[Nick Cox <njcoxstata@gmail.com>]

You could parse your identifiers using -substr()- to split out parts.
I've found many times that people underestimate the possibilities of
the very simplest string functions. There is a tutorial on functions
often neglected in a 2011 paper

http://www.stata-journal.com/article.html?article=dm0058

Or you could use regular expression tools. -moss- from SSC could work
with examples like this:

. l

     +-------------+
     |          id |
     |-------------|
  1. | BTG09A00001 |
  2. | BTG10A00001 |
  3. | BGM09A00027 |
  4. | BGM10A00027 |
     +-------------+

. moss id, match("([0-9]+)") regex prefix(n_)

. moss id, match("([A-Z]+)") regex prefix(a_)

. l id *match*

     +---------------------------------------------------------+
     |          id   n_match1   n_match2   a_match1   a_match2 |
     |---------------------------------------------------------|
  1. | BTG09A00001         09      00001        BTG          A |
  2. | BTG10A00001         10      00001        BTG          A |
  3. | BGM09A00027         09      00027        BGM          A |
  4. | BGM10A00027         10      00027        BGM          A |
     +---------------------------------------------------------+

That's split the identifiers into alphabetic and numeric sequences. I
took your examples literally in producing these commands. In your
case, you don't care about the result of -a_match2- but I left in
above to show that -moss- can split out two or more components, not
just one as is typically of calls to -substr()-.

That said, Stata makes it easier to create identifiers that will work
well across Stata's commands. Do-it-yourself identifiers can just make
tables and graphs unwieldy.

For a 2007 review, see

http://www.stata-journal.com/article.html?article=dm0034

The .pdf for that is accessible to all at

http://www.stata-journal.com/sjpdf.html?articlenum=dm0034

Nick

On Thu, Feb 7, 2013 at 1:37 PM, Michler, Jeffrey D <jmichler@purdue.edu> wrote:

> I have a dataset which includes household ID variables in an alphanumeric format. The letters are abbreviations of the village a household comes from.  In addition to being in an alphanumeric format, the HH ID has a year element so that the HH ID for 2010 is slightly different than it was for 2009.  I am looking to convert the alphanumeric HH id into a unique id for constructing a panel. I need to replace the 3 letter village abbreviations with a 3 digit number plus I need to drop the year id.
>
> An example may clarify. Right now HH IDs look like BTG09A00001, BTG10A00001, BGM09A00027, BGM10A00027.
>
> I want to replace the village code (BTG, BGM) with a numerical sequence. I also want to drop the year sequence (09, 10) so that HH ID is consistent for the HH across years, and I want to drop the A, which plays to role in my dataset. Ideally, this would compress the 4 HH ID I gave as examples into just 2 IDs that would look like 10100001 and 10200027.

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