Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: Using regexr() in local macros without using an = sign


From   Nick Cox <[email protected]>
To   [email protected]
Subject   Re: st: Using regexr() in local macros without using an = sign
Date   Thu, 29 Nov 2012 10:13:16 +0000

You should look at -help macrolists-. (More generally, if you start
browsing at -help macro- you will find several tools that you can
use.)

. local all "aname bname cname"

. local b "bname"

. local allbut : list all - b

. mac li
<snip>
_allbut:        aname cname
_b:             bname
_all:           aname bname cname

You'd need to call that in a loop in your case, but you are presumably
doing that already.

I note also that you use

if strmatch("`v'", "`i'") == 0

when

if "`v'" != "`i'"

is simpler, although as above the macrolist approach is better.

If you are looking for the characters "b." in a name then

foreach v of local varlist {
       if strpos("`v'","b.") local onlybases `onlybases' `v'
      else local nobases `nobases' `v'
}

is an alternative.

Nick

On Thu, Nov 29, 2012 at 8:56 AM, Abe N <[email protected]> wrote:

> Actually, I had to modify that last part in the loop to the following:
>    local elrest
>    foreach v of local rest {
>      if strmatch("`v'", "``i''")==0 local elrest `elrest' `v'
>    }
> and it seems to be working thus far.  Going to let this run overnight to
> test and see if it works out tomorrow.  THanks again!

On Thu, 11/29/12, Abe N <[email protected]> wrote:

>> Thanks for the help Nick, the code
>> you showed me is so much simpler than what I was
>> trying.  It seems to work, although I tweaked it a
>> little bit, because I think it's more important that the
>> variable have "b." in there somewhere.  Here's the code
>> I ended up using in case anyone else runs into this:
>>
>>     foreach v of local varlist {
>>       if strmatch("`v'","*b.*") == 1 local
>> onlybases `onlybases' `v'
>>       else local nobases `nobases' `v'
>>     }
>>
>> I've now run into a slightly different issue that maybe you
>> can help me with.  There is somewhere else in my
>> program that I've evaluated a macro and again I run into the
>> character limit.  If you were to take the variables
>> contained in onlybases from above, I'm running a loop where
>> first for each variable I run a logistic by itself, then I
>> run a logistic with all the variables in `onlybases' except
>> without the variable.
>>
>> So first for each variable in `onlybases', I perform the
>> following commands (``i'' is the variable here, used after
>> tokenizing and the loop is done using a while command)
>>   while "``i''"~="" {
>>    logistic `dv' ``i'' if `touse'
>>    local elrest = regexr("`rest'", "``i''",
>> "")
>>    logistic `dv' `elrest' `onlybases' if
>> `touse'
>>   }
>>
>> Is there a way I could accomplish this without evaluating
>> again with an =?  Now that I think about it, maybe I
>> should try repeating something like the above within this
>> loop?  Maybe:
>> foreach v of local rest {
>>   if strmatch("`rest'", ``i'')==0 local elrest `elrest'
>> `v'
>> }
>> I think that would but all the variables except ``i'' in a
>> local, right?

On Wed, 11/28/12, Nick Cox <[email protected]> wrote:

>> > If I understand your problem, a more basic approach is
>> > something like
>> >
>> >      fvexpand `varlist'
>> >      local varlist "`r(varlist)'"
>> >      foreach v of local varlist {
>> >
>> >    if strpos("`v'", 1) == "0" local sheep
>> > `sheep' `v'
>> >
>> >    else local goats `goats' `v'
>> >     }
>> >
>> > You can use regular expressions if you like, but the
>> key
>> > issue appears
>> > to be whether "0" is the first character.
>> >
>> > -tokenize- is not going to help much, if at all. You
>> want to
>> > look at
>> > each name in a list and whether the names are in
>> individual
>> > macros or
>> > just one long macro is not the main question.
>> >
>> > -regexr()- is a function, not a command.

On Wed, Nov 28, 2012 at 4:55 AM, Abe N <[email protected] wrote:

>> > I've been trying to write a program that at the basics
>> of it
>> > runs a series of logistic regressions and while I got
>> it to
>> > work for the most part, I've been trying to do
>> something
>> > specifically with factor variables (I'm using version
>> 11.2
>> > STATA/SE)
>> > >
>> > > What I'm trying to do currently I think is pretty
>> > basic.  I'm starting with a variable list that has
>> > factor variables (ex. my current varlist will
>> basically
>> > contain: dependent_variable i.age_cat i.year female),
>> > expanding it, and then trying to create two new macros,
>> one
>> > that contains the base variables (ex. 0b.age_cat
>> 1998b.year)
>> > and another that contains the rest (ex. died 1.age_cat
>> > 2.age_cat 1999.year 2000.year 2001.year female).
>> > >
>> > > The way I went about this at first worked well for
>> a
>> > couple variables, but I was running into the issue of
>> > character limits when you evaluate a macro with an =
>> > sign.  Here's the code I was using:
>> > >     fvexpand `varlist'
>> > >     local varlist = "`r(varlist)'"
>> > >     local nobases = "`varlist'"
>> > >     local loopvar =
>> > regexm("`nobases'", "[a-zA-Z0-9]+[b][\.][a-zA-Z0-9_]+[
>> ]*")
>> > >     local onlybases
>> > >     while "`loopvar'"=="1" {
>> > >       local onlybases =
>> > "`onlybases' "+regexs(0)
>> > >       local nobases =
>> > regexr("`nobases'", "[a-zA-Z0-9]+[b][\.][a-zA-Z0-9_]+[
>> ]*",
>> > "")
>> > >       local loopvar =
>> > regexm("`nobases'", "[a-zA-Z0-9]+[b][\.][a-zA-Z0-9_]+[
>> ]*")
>> > >     }
>> > >     local varlist "`nobases'"
>> > > Then when I run logistics later, I would
>> basically
>> > use:
>> > >     logistic `nobases' `onlybases'
>> > >
>> > > The problem with this method is the macro
>> `nobases'
>> > will cut off after a certain # of characters.  So I
>> > tried doing things without = signs, but I've been
>> running
>> > into a lot of problems because I'm using the regexr()
>> > command and it doesn't seem to like me using that
>> without an
>> > = sign.
>> > >
>> > > Now that I've written this up, I'm wondering if
>> this
>> > would have been a lot easier using tokenize, but I
>> think I
>> > avoided it because I use it later for a very key aspect
>> of
>> > the program and I think I was worried it might mess
>> with
>> > some of my later loops.
*
*   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/


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index