Re: [Gretl-devel] Proposal for function package cleanup
Sorry, I am not on the gretl-devel list. I will join later.
>>So does this mean that fcModels is the more general one and
that
HoltWinters is redundant?
No, they are not redundant. My fcModels.gfn could partially be thought to
extend parts of Ignacio's HoltWinters despite I also wrote other
forecasting models.
What Ignacio said, "...An important difference is that he (Yang) is looking
for 'optimal' smoothing parameters", is just what I did in the
fcModels.gfn. Why I included Ignacio's HoltWinters is just as what Ignacio
said, "currently we don't have a way to call a function from a different
package in the current gretl functions. So I just directly included the
original script of Ignacio's.
Actually I would be very glad to see in the future that we would have such
a way to call a function from a different package so that I don't have to
rewrite the same functions with existing ones. Without this kind of
"inclusion", it turned out the reason why fcModels.gfn has not updated HW2
with the ultimate versions of HoltWinters. (I personally apologize about
this.)
Of course, it will be my pleasure to cooperate with Ignacio. But in fact,
it was not my intention to write another version of Holt and Winters'. If
Ignacio's package can be called, I will tend to call his existing ones.
>>could you edit your package to make it English-only
Though I personally think currently 'English-only' is not very good for me
to promote gretl in Taiwan as well as for Chinese users, I surely will
conform to the decision of the gretl society. I support sven's idea that it
is a clear example of how we need a way to translate function packages.
I agree to remove HW2 from my fcModels.gfn package and make it English-only
but please allow me to do this after ending of this semester since I will
have an "online" final exam in which fcModels.gfn has to be used by
students.
Thanks for your great contributions.
Yi-Nung Yang
Associate Professor
Dept. of International Business
Chung Yuan Christian University, Taiwan.
2015-06-19 20:01 GMT+08:00 Ignacio Diaz-Emparanza <
ignacio.diaz-emparanza(a)ehu.eus>:
El 18/06/15 a las 09:09, Sven Schreiber escribió:
> (Yi-Nung [on cc:], are you on the gretl-devel list? If not, would it be
> possible to join, since you are a package author? Thanks. If you reply,
> could you keep the devel-list address on cc: please [reply-to-all].)
>
> Am 16.06.2015 um 19:32 schrieb Ignacio Diaz-Emparanza:
>
>> El 16/06/15 a las 18:49, Sven Schreiber escribió:
>>
>>> 5) fcModels / HoltWinters (Yang / Diaz) -- Again, cooperation would be
>>> good. But both packages were updated relatively recently, which is
>>> also a good sign. Don't know what else to do with this.
>>>
>> With respect to this, I see that Yi-Nung, by one side is nesting my
>> package with their other ones, what I think is very logical.
>>
> So does this mean that fcModels is the more general one and that
> HoltWinters is redundant? Or would this be a case for different
> implementations that are both useful?
>
I have just taken a look to 'fcmodels.gfn'. Yi-Nung has made a broader
package which includes between his functions Holt and Winters filters
separately, and probably from a slightly different formulas than mine. Then
included my HoltWinters function as an alternative (a new public function,
HW2) in his package. An important difference is that he is looking for
'optimal' smoothing parameters and in my function these are fixed. Their
functions and HW2 produce different results.
Furthermore it seems he has not updated HW2 with the ultimate versions of
my HoltWinters package given that they are also producing different results.
I think the more simple and logical position now (given that currently we
don't have a way to call a function from a different package) is to ask him
to remove HW2 from his package. Then 'fcmodels.gfn' and HoltWinters.gfn
would have two different implementations of the same thing. May be in a
future we can cooperate and convert them in a unique package (but I have
not much time for this right now).
--
Ignacio Díaz-Emparanza
Departamento de Economía Aplicada III (Econometría y Estadística)
Universidad del País Vasco - Euskalherriko Unibertsitatea, UPV/EHU
Tfno: (+34) 94 601 3732
http://www.ehu.eus/ea3
Attachments:
- attachment.html (text/html — 5.6 KB)