1:56 p.m.
Dear Gretl Community,
Is there any way to retrieve the roots of an ARIMA model? Is there an
accessor that allows this kind of operation?
My aim goal is to build an automatic ARIMA modelling script that saves only
the models with signicant parameters and roots outside the unit circle.
Many thanks,
Henrique Andrade
3:06 p.m.
Henrique,
I think you could write the ARMA model in state-space form and
computing eigenvalues of that. For example, suppose that you have a
series y, modeling as AR(3)
arima 3 0 0; y
b=$coeff
c=I(2)~zeros(2,1)
A=b' | c
d=eigenvalue(A)
In this case you need that elements in d must be inside unit circle.
Rodrigo.
2012/1/31 Henrique Andrade <henrique.coelho(a)gmail.com>:
Dear Gretl Community,
Is there any way to retrieve the roots of an ARIMA model? Is there an
accessor that allows this kind of operation?
My aim goal is to build an automatic ARIMA modelling script that saves only
the models with signicant parameters and roots outside the unit circle.
Many thanks,
Henrique Andrade
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
3:13 p.m.
On Tue, 31 Jan 2012, Henrique Andrade wrote:
Is there any way to retrieve the roots of an ARIMA model? Is there
an
accessor that allows this kind of operation?
They are attached to the model in question under the name
"roots". While there is no "$roots" accessor as such, it's now
possible to retrieve them (in CVS and snapshots) via the
special form $["roots"]. This form will work for any named
data attached to the last model in the form of a scalar,
series or matrix. Example:
open data9-7
arma 1 2 ; QNC
matrix r = $["roots"]
r
This form treats the last model as if it were a "bundle", from
which any suitable named data can be extracted.
Allin Cottrell
2:31 p.m.
On 01/31/2012 09:13 PM, Allin Cottrell wrote:
On Tue, 31 Jan 2012, Henrique Andrade wrote:
> Is there any way to retrieve the roots of an ARIMA model? Is there an
> accessor that allows this kind of operation?
They are attached to the model in question under the name
"roots". While there is no "$roots" accessor as such, it's now
possible to retrieve them (in CVS and snapshots) via the
special form $["roots"]. This form will work for any named
data attached to the last model in the form of a scalar,
series or matrix. Example:
open data9-7
arma 1 2 ; QNC
matrix r = $["roots"]
r
This form treats the last model as if it were a "bundle", from
which any suitable named data can be extracted.
Ho ho -- hold it :-)
Seriously, isn't this new syntax being introduced a little hastily? So
far everything that could be accessed got its $-accessor. This $[] thing
seems to be all new. I'm not saying it's bad, but it seems to exist in
parallel to the old-style accessor. After all, all existing accessors
could now be re-defined, for example $coeff -> $["coeff"]. Maybe that's
the way to go (old-style accessors could be kept as syntactic sugar for
backward-compatibility).
cheers,
sven
2:56 p.m.
On Wed, 1 Feb 2012, Sven Schreiber wrote:
On 01/31/2012 09:13 PM, Allin Cottrell wrote:
> On Tue, 31 Jan 2012, Henrique Andrade wrote:
>
>> Is there any way to retrieve the roots of an ARIMA model? Is there an
>> accessor that allows this kind of operation?
>
> They are attached to the model in question under the name
> "roots". While there is no "$roots" accessor as such, it's
now
> possible to retrieve them (in CVS and snapshots) via the
> special form $["roots"]. This form will work for any named
> data attached to the last model in the form of a scalar,
> series or matrix. Example:
>
> open data9-7
> arma 1 2 ; QNC
> matrix r = $["roots"]
> r
>
> This form treats the last model as if it were a "bundle", from
> which any suitable named data can be extracted.
Ho ho -- hold it :-)
Seriously, isn't this new syntax being introduced a little hastily? So
far everything that could be accessed got its $-accessor. This $[] thing
seems to be all new. I'm not saying it's bad, but it seems to exist in
parallel to the old-style accessor. After all, all existing accessors
could now be re-defined, for example $coeff -> $["coeff"]. Maybe that's
the way to go (old-style accessors could be kept as syntactic sugar for
backward-compatibility).
Though I mentioned it here, the $["name"] thing isn't really
ready for roll-out. To see what's available in that mode you
really have to grep the gretl sources for
gretl_model_set_data
gretl_model_set_double
gretl_model_set_int
and so on. These "set" functions are ways of attaching named
data for internal use, but the bundle-key syntax makes it easy
for users to grab a copy -- if you know what to look for.
Allin
8:58 p.m.
Hi to all,
Folks I'm a little rusty...
What's now being meant by the 'roots' of an ARIMA model, please?
Best wishes
Richard Hudson
Dr RJF Hudson Qld Australia
rjfhud(a)powerup.com.au
----- Original Message -----
From: Sven Schreiber
To: gretl-users(a)lists.wfu.edu
Sent: Thursday, February 02, 2012 5:31 AM
Subject: Re: [Gretl-users] Retrieving the roots of an ARIMA
On 01/31/2012 09:13 PM, Allin Cottrell wrote:
On Tue, 31 Jan 2012, Henrique Andrade wrote:
> Is there any way to retrieve the roots of an ARIMA model? Is there an
> accessor that allows this kind of operation?
They are attached to the model in question under the name
"roots". While there is no "$roots" accessor as such, it's now
possible to retrieve them (in CVS and snapshots) via the
special form $["roots"]. This form will work for any named
data attached to the last model in the form of a scalar,
series or matrix. Example:
open data9-7
arma 1 2 ; QNC
matrix r = $["roots"]
r
This form treats the last model as if it were a "bundle", from
which any suitable named data can be extracted.
Ho ho -- hold it :-)
Seriously, isn't this new syntax being introduced a little hastily? So
far everything that could be accessed got its $-accessor. This $[] thing
seems to be all new. I'm not saying it's bad, but it seems to exist in
parallel to the old-style accessor. After all, all existing accessors
could now be re-defined, for example $coeff -> $["coeff"]. Maybe
that's
the way to go (old-style accessors could be kept as syntactic sugar for
backward-compatibility).
cheers,
sven
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
4:26 a.m.
El 02/02/12 02:58, Dr RJF Hudson escribió:
Hi to all,
Folks I'm a little rusty...
What's now being meant by the 'roots' of an ARIMA model, please?
Best wishes
Richard Hudson
Dr RJF Hudson Qld Australia
The roots of the AR and MA polynomials in the lag operator. They are at
the end of the model output that gretl produces. And Henrique was asking
for a way to save them for subsequent use in a script.
--
Ignacio Diaz-Emparanza
DEPARTAMENTO DE ECONOMÍA APLICADA III (ECONOMETRÍA Y ESTADÍSTICA)
UPV/EHU Avda. Lehendakari Aguirre, 83 | 48015 BILBAO
T.: +34 946013732 | F.: +34 946013754
www.ea3.ehu.es
6:05 p.m.
New subject: Copy and Pasting Scripts on Grelt Console
Hi All,
Could someone gist me if there's a means of copying and pasting scripts on
the Grelt console? Indeed the monotony of retyping the same script on and again makes me
sick.
5:34 a.m.
New subject: Copy and Pasting Scripts on Grelt Console
On 02/05/2012 12:05 AM, Anutechia Asongu wrote:
Hi All,
Could someone gist me if there's a means of copying and pasting
scripts on the Grelt console? Indeed the monotony of retyping the same
script on and again makes me sick.
?? Are you aware that a gretl script can/should be run in its entirety
(for example by clicking the execute button in the script window) and
not line-by-line in the console?
(Actually, one may even live without the console, because in a script
window you can select parts --e.g. a single line-- and by right-clicking
execute only the selected part. But that's a different topic.)
hth,
sven
12:37 p.m.
New subject: Copy and Pasting Scripts on Grelt Console
Hi Sven,
Many thanks for the great service. Scripting is much easier than I had ever
immagined
________________________________
From: Sven Schreiber <svetosch(a)gmx.net>
To: gretl-users(a)lists.wfu.edu
Sent: Sunday, February 5, 2012 11:34 AM
Subject: Re: [Gretl-users] Copy and Pasting Scripts on Grelt Console
On 02/05/2012 12:05 AM, Anutechia Asongu wrote:
Hi All,
Could someone gist me if there's a means of copying and pasting
scripts on the Grelt console? Indeed the monotony of retyping the same
script on and again makes me sick.
?? Are you aware that a gretl script can/should be run in its entirety
(for example by clicking the execute button in the script window) and
not line-by-line in the console?
(Actually, one may even live without the console, because in a script
window you can select parts --e.g. a single line-- and by right-clicking
execute only the selected part. But that's a different topic.)
hth,
sven
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
4489
days inactive
4494
days old
9 comments
7 participants
participants (7)
-
Allin Cottrell -
Anutechia Asongu -
Dr RJF Hudson -
Henrique Andrade -
Ignacio Diaz-Emparanza -
Rodrigo Alfaro Arancibia -
Sven Schreiber