NJC can offer a precise answer, but my take would be
gen byte one = 1
bys group: gen varmean = sum(mean)/sum(one)
by group: keep if _n==_N
keep whatever
Topics like those should've been covered somewhere in Nick's column in
Stata Journal, or in Stata tips. -egen- is slow as it does a lot of
checks and parsing and stuff -- for big processing jobs, single-liners
like above are always notably faster. -collapse- should be at least a
tad faster than -egen-, but again I would expect it to lose to the
above code.
On Fri, Apr 25, 2008 at 2:37 PM, Jeph Herrin <[email protected]> wrote:
>
> I'm optimizing some code that needs to run often
> for a simulation, and am wondering if I should
> expect any difference in processing time between
>
> bys group: egen varmean=mean(myvar)
> bys group: keep if _n==1
> keep group varmean
>
> and
>
> collapse (mean) varmean=myvar, by(group)
>
> and if so, which would be faster?
>
> I know I could run some tests myself, but figured
> that others had either already done so or at least
> would have some insight.
>
--
Stas Kolenikov, also found at http://stas.kolenikov.name
Small print: Please do not reply to my Gmail address as I don't check
it regularly.
*
* 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/
