[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
RE: st: AW: tsset on subsample of observations
I think Tom is right. The short answer is No;
a longer answer is Yes, by sweeping problematic
observations out of the way.
First, note the syntax: there is no support for
-if- and -in- with -tsset-.
Second, note the logic: -tsset- is for making
declarations about your dataset. I know you
want to make declarations about part of your
dataset, but that is not supported.
I can't see a way of doing what you want
except by sweeping repeated observations
out of the way, at least temporarily, via
-drop- or -keep-.
> Hi Tom,
> Thanks for the suggestion, however it doesn't really get at
> what I want,...
> After defining tsset, I want to create variables like
> gen z = x-l.x
> and keep this variables in the dataset. For the duplicate obs. z
> should take missing values.
> By using -preserve- / -restore- I lose z after restoring the
> old dataset.
> Any other suggestion will be greatly appreciated
On 11/11/06, Thomas Erdmann <firstname.lastname@example.org> wrote:
> > Couldnt you work with -preserve- and -restore- ?
> > duplicates tag x y, generate(dup)
> > preserve
> > drop if dup~=0
> > tsset x y
> > ...
> > restore
> > Wouldn't it be better to keep one of the duplicate observations?
> > Duplicates drop, force
> > Instead of
> > Drop if dup~=0
> > Is there a way to tsset a subsample of my observations?
> > I want to perform operations that involve time series
> commands, but if
> > I consider my entire dataset I have repeated time values within the
> > panel.
> > I would like to Stata to tsset my observations for that
> subset of obs.
> > that are uniquely identified within each panel... something like:
> > tsset x y if unique==1
> > Is there a way to do this?
> > Otherwise I would be forced to drop the repeated observations, but
> > since I need those obs. for other exercises this will force
> me to save
> > datasets: one with duplicates and another one without them.
> > that seems highly inefficient.
* For searches and help try: