Creating X-12-ARIMA spec file
by 不提供 不提供
Dear all:
I have a time series data in gretl. Can I convert it to x-12-arima spec file?
Because it can not creat X-12-ARIMA spec file with excel.
I do not know how to creat X-12-ARIMA spec when I have data in excel,
(maybe convert excel to dat. but I dont know how to do this)
So I want to creat spec through gretl.
Thanks a lot
13 years, 3 months
About gretl with the Traditional Chinese version (zh_TW)
by yinung＠Gmail
Dear Allin:
I know that the default fixed font of the current gretl version for windows
with Traditional Chinese is set to "MS Gothic"
Would this setting be changed to "NSimsun," because I found that it is
better than "MS Gothic"
Best regards,
Yi-Nung
13 years, 3 months
Re: [Gretl-users] VECM alpha restrictions
by Allin Cottrell
On Tue, 11 Jan 2011, Riccardo (Jack) Lucchetti wrote:
> On Tue, 11 Jan 2011, Allin Cottrell wrote:
>
> > Part of my motivation for the switch was that if you give a
> > restriction such as "a[1] = 0" (with a single index) this means
> > that all EC coefficients in equation 1 are restricted to zero, or
> > variable 1 is weakly exogenous. And it seemed more intuitive that
> > if you give just one index it should be read as the leading one,
> > not the trailing one.
>
> This makes good sense, but does it warrant a compatibility breakage? I'm
> with Sven on this one.
OK, thanks, time to revert! This will go into CVS before long.
Allin
13 years, 3 months
Re: [Gretl-users] VECM alpha restrictions
by Allin Cottrell
On Tue, 11 Jan 2011, Sven Schreiber wrote:
> I'm not sure [gretl's previous indexation of \alpha
> restrictions] was a bug. Of course it's all just a matter of
> convention, but given that for beta-indexing in this case we
> also had the CI relation number as the leading index, there's a
> case to do that for alpha, too, which was the old behavior.
OK, maybe I was wrong. We could revert my change and add an
explicit note to the manual regarding the indexation convention
used. Anyone else have thoughts on this?
Part of my motivation for the switch was that if you give a
restriction such as "a[1] = 0" (with a single index) this means
that all EC coefficients in equation 1 are restricted to zero, or
variable 1 is weakly exogenous. And it seemed more intuitive that
if you give just one index it should be read as the leading one,
not the trailing one. However, I take your point that this is a
matter of convention; all that really matters is that the manual
is clear on what we're doing.
(Either way, I don't want to mess with our existing \beta
indexation at this point.)
Allin
13 years, 3 months
Re: [Gretl-users] VECM alpha restrictions
by Allin Cottrell
On Tue, 11 Jan 2011, Artur Tarassow wrote:
> I also prefer the old indexation-style even though I understand your
> argument of
> "a[1] = 0" (with a single index), Allin. But I am in line with Sven that
> it is better to have the same approach for alpha and beta restriction
> (CI relation number as the leading index).
>
> Also the new indexation would affect all our scripts which we have
> written yet. ;-)
Thanks. I'll wait a little to see if we hear from anyone else, but
unless we get a strong vote to the contrary I'll revert to our
old \alpha indexation.
Allin
13 years, 3 months
VECM alpha restrictions
by Allin Cottrell
Hello all,
Following from a bug report from Artur T, I've been examining the
handling of restrictions on the \alpha matrix in a VECM. Among
other things, I noticed that we've had the interpretation of the
row and column indices reversed, in the context of a restriction
such as
a[3,1] = 0
That is, we've read this as EC-term 3, equation 1, which is in
fact \alpha_{1,3}. This probably happened because for internal
purposes we represent the restriction in terms of a vector that
gets multiplied into vec(\alpha'). Nonetheless, it seems
like a definite bug to me, and it's now corrected in CVS (and
duly noted in gretl's backward-incompatibility log).
>From now on an expression such as "a[3,1]" in a VECM restriction
will be read as specifying \alpha_{3,1}, the coefficient on the
first EC term in the third equation.
Allin Cottrell
13 years, 3 months
How to generate forecasts
by 不提供 不提供
Dear all:
I choose the monthly (or quarterly) data from1990 to 2008 for in-sample forecasting, 2009 to 2010 for out-of-sample forecasting. Finally, I find the ARIMA(1,1,1)(2,1,1) is the most appropriate model, then I want to use this model to forecast for 2011-2013.
Can it generate forecasts for the future in Gretl? How to do this?
Thanks a lot
13 years, 3 months
Some questions about X-12-ARIMA
by 不提供 不提供
Dear all:
Allin Cottrell, thank you.
When I choose the options of Model/Time series/ARIMA/Using X-12-ARIMA to run the X-12-ARIMA model,
are the explanatory variables such as as number of working days,day of week effects, length of month etc inclued already?
If they are included automatically,
maybe there should be differences between native-gretl and X-12-ARIMA.
This question still puzzles me?
Thanks a lot
13 years, 3 months
Re: [Gretl-users] Some questions about X-12-ARIMA
by Allin Cottrell
On Mon, 10 Jan 2011, [big5] ������ ������ wrote:
> When I choose the options of Model/Time series/ARIMA/Using
> X-12-ARIMA to run the X-12-ARIMA model, are the explanatory
> variables such as as number of working days,day of week effects,
> length of month etc inclued already?
No, none of these features are included automatically when you use
X-12-ARIMA from the ARIMA menu. The idea is to produce output that
is as closely compatible as possible with gretl's native ARIMA.
In gretl, you also have access to X-12-ARIMA under the /Variable
menu: "X-12-ARIMA analysis". In this context you have the option
of engaging correction for outliers and Trading Day effects.
If you want greater control over X-12-ARIMA you'd have to write
your own .spc file. One modification that might be useful, and
that could be implemented without great difficulty, would be to
give the gretl user the opportunity to edit the spc file from
within gretl before sending it to X-12-ARIMA.
Allin Cottrell
13 years, 3 months
Some questions about X-12-ARIMA
by 不提供 不提供
Dear all:
Allin Cottrell, Dr RJF Hudson & John C. Frain, Thank you.
Maybe my questions were not expressed precisely. I make it clearer here.
1. In your examples you are estimating the same seasonal ARIMA model usingthe same method (conditional ML) (a) via gretl itself and (b) via
X-12-ARIMA.
(Allin Cottrell)
1.So the X-12-ARIMA(1,1,1)(1,1,0) model I get is not
the RegARIMA(1,1,1)(1,1,0) (RegARIMA defined by X-12-ARIMA – Reference Manual, Version 0.3. U.S. Census Bureau).
Is it?
The options of Model/Time series/ARIMA/Using X-12-ARIMA in gretl 1.9.1 cvs doesn't mean running RegARIMA,
it doesn't include the variables such as as number of working days, day of week effects, length of month etc .Is it?
2. How do I run RegARIMA in gretl 1.9.1 cvs?
How do I input the variables such as number of working days, day of week effects, length of month etc
as X-12-ARIMA – Reference Manual, Version 0.3. (U.S. Census Bureau) defining and show their outcome?
3. If the X-12-ARIMA(1,1,1)(1,1,0) model I get is RegARIMA(1,1,1)(1,1,0), so it includes the variables above.
Maybe the outcomes of ARIMA(1,1,1)(1,1,0) and X-12-ARIMA(1,1,1)(1,1,0) should be different, not the same.(I dont know)
In a word, Can I run RegARIMA in Gretl 1.9.1 cvs (Windows)?
Does the options of Model/Time series/ARIMA/Using X-12-ARIMA in gretl 1.9.1 cvs mean running RegARIMA?
Thanks a lot
13 years, 3 months
