Re: [Gretl-users] bug in gap_filler() (extra v0.5)
Am 18.12.18 um 23:19 schrieb Riccardo (Jack) Lucchetti:
On Tue, 18 Dec 2018, Artur T. wrote:
> Hi all,
>
> I think I've found a bug in the gap_filler() function which is part of
> the extra package. This happens on current git using Ubuntu.
[...]
That's by design. The function (as its name suggests) is meant to fill
gaps by joining valid observations somehow. If you have suggestions on
what to do if NAs occur at the beginning or the end of the sample,
they'd be most welcome.
Ok, this is what I expected after having identified this "behavior".
Nevertheless, I was surprised to see this happen. Of course one can't
interpolate the initial or last obs. However, I guess the user usually
expects that this corner-issue is somehow solved for him/her or at least
a warning is returned.
Actually, I would have expected that the NA at the very end is replaced
by the last valid value or so. But fair enough: endless options are
thinkable...
Maybe somebody else comes up with a useful idea.
Artur