Statalist


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

Re: st: How to loop only over the periods in which the commands are feasible, and automatically ignore the others?


From   Stata Chris <statachris@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: How to loop only over the periods in which the commands are feasible, and automatically ignore the others?
Date   Tue, 4 Aug 2009 11:10:54 -0400

Thanks, I think the "Capture" command was exactly what I had been looking for!

As for (2), I had suspected that forval might require explicit values,
but was not sure whether this was indeed the reason for the error, so
it's good to know this for next time now!

All best,
Chris

2009/8/4 Nick Cox <n.j.cox@stata.com>:
> One short answer to (1) is that you can put -capture- around a block of
> statements that might fail.
>
> The code might look like this
>
> forvalues i=458/515 {
>        capture {
>                reg delta L.delta if period==`i'
>                replace beta = adjust *  _b[L.delta] if period==`i'
>                ...
>        }
> }
>
> Another approach is something like
>
> forvalues i = 458/515 {
>        qui count if period == `i' & delta != .
>        if r(N) > 0 {
>                ...
>        }
> }
>
> where both tests
>
>        if period == `i' & delta != .
>        if r(N) > 0
>
> should be replaced by any test more appropriate.
>
> In terms of (2), the most obvious error is
>
> forvalues i=period2_min(1)period2_max{
>
> as -forval- requires explicit values. That said, getting the approach in (2)
> exactly right is in my experience harder work than other solutions sketched
> above.
>
> Nick
>
> Stata Chris wrote:
>
>> (1)
>> I wanted to run the loop below, where for some of the periods either
>> delta or L.delta does not exist. Now it seems that with the code
>> below, rather than executing the commands for the periods where the
>> necessary observations do exist and ignoring the others, what happens
>> is that whenever at least for one period the command cannot be
>> executed Stata gets stuck entirely. Is there a smart way to tell it to
>> just ignore those periods and continue with the next one?
>>
>> . gen beta         = .
>> . gen se_beta      = .
>> . gen t_beta       = .
>> . forvalues i=458(1)515 {
>>     . reg delta L.delta                                      if
>> period==`i'
>>     . replace beta        = adjust *  _b[L.delta]    if period==`i'
>>     . replace se_beta   = adjust * _se[L.delta]   if period==`i'
>>     . replace t_beta     = beta / se_beta            if period==`i'
>> . }
>>
>> (2) What I tried is to kick out all non-feasible periods before
>> starting the loop, and creating a new period2-variable that contains
>> only those periods in which it is feasible. I tried this with the code
>> below, but here Stata tells me that the syntax is invalid where I
>> would like it to start the loop...
>>
>> . save temp1.dta, replace
>> . sort obs_fundasset period
>> . keep if delta!=. & L.delta!=.
>> . duplicates drop period, force
>> . gen period2=_n
>> . save temp2.dta, replace
>> . clear all
>> . use temp1.dta
>> . merge obs_triple using temp2, sort
>> . summarize period2
>> . scalar period2_min = r(min)
>> . scalar period2_max = r(max)
>> . forvalues i=period2_min(1)period2_max{
>>     . reg delta L.delta                               if period2==`i'
>>     . replace beta          = adjust *  _b[L.delta]   if period2==`i'
>>     . replace se_beta       = adjust * _se[L.delta]   if period2==`i'
>>     . replace t_beta        = beta / se_beta          if period2==`i'
>> . }
>>
>
> *
> *   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index