Pleased to hear -levelsof- is useful here.
Nick
[email protected]
Steichen, Thomas J.
> A variation on "levelsof id if <condition>" appears to
> accomplish what I need.
>
> It is far more efficient with respect to programmer time and
> effort but I haven't checked whether it is cpu-cycle efficient.
Nick Cox
> > -findobs- by Sean Becketti is the most substantial
> > work in this area that springs to mind.
> >
> > I can't suggest anything much simpler than
> >
> > gen long id = _n
> >
> > su id if <condition>, meanonly
> >
> > list id if <condition>
> >
> > levelsof id if <condition>
Steichen, Thomas J.
> > > I need to efficiently capture, for example, the record id
> > > where something is true about that record.
> > >
> > > For example, if I have the following made-up data:
> > >
> > > id x
> > >
> > > a 6
> > > b 32
> > > c 6
> > > d 11
> > >
> > > I may want to capture a record id where x = max(x).
> > > I know I can get it via:
> > >
> > > . egen mx = max(x)
> > > . gen rx = 1 if x == mx
> > > . sort rx
> > > . local id_info = id (or, more accurately, id[1])
> > > . di "ID info: `id_info'"
> > > ID info: b
> > >
> > > But an egen, gen, and sort seems to be an expensive way to do
> > > this with a large dataset.
> > >
> > > The above solution will, of course, only capture one record
> > > id, where x=max(x).
> > >
> > > If I had wanted x=min(x), it would be nice if id_info
> > > contained "a c". This is doable also, but becomes even more
> > > expensive in
> > > computing time.
> > >
> > > Can someone propose a more efficient way or point out a
> > > built-in command that I've missed?
*
* 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/
