Re: [Gretl-devel] remove missings permanently destroys time series structure
On Fri, 1 Mar 2019, Sven Schreiber wrote:
Am 01.03.2019 um 20:27 schrieb Allin Cottrell:
> On Fri, 1 Mar 2019, Sven Schreiber wrote:
>
>> Hi,
>> tried to throw out all obs with missings (one series only in a time series
>> dataset), and when ticking "make permanent" I got an undated dataset.
>>
>> This is with 2019a.
>
> Sven, could you give an idea of what you expected here? Seems to me that
> while discarding a series containing NAs would leave you with a time series
> dataset, discarding observations is bound to destroy the time series
> property (unless the NAs were just at the start and/or end of the data).
That's what I meant, an "interior" sample with valid observations and then
some junk portions of NA before and after the valid block. At least that's
what I presumed, maybe I have to check...
OK, on a little simulation I see that case is not being handled
correctly -- i.e. time series structure not preserved on making the
restriction permanent. Fix should be forthcoming before long.
Allin