11:28 a.m.
Dear all,
It seems the list now
is more powerful with attachments
so 792 M4 economic monthly series
as .gdt file attached
Oleh
1:28 p.m.
On Tue, 20 Mar 2018, oleg_komashko(a)ukr.net wrote:
Dear all,
It seems the list now
is more powerful with attachments
so 792 M4 economic monthly series
as .gdt file attached
I'm attaching a script file that performs some performance comparisons
between as197 and x12; it uses some stock series, some simulated series
and series from Oleh's dataset. Have fun.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
3:47 p.m.
On Tue, 20 Mar 2018, Riccardo (Jack) Lucchetti wrote:
On Tue, 20 Mar 2018, oleg_komashko(a)ukr.net wrote:
> Dear all,
> It seems the list now
> is more powerful with attachments
> so 792 M4 economic monthly series
> as .gdt file attached
I'm attaching a script file that performs some performance comparisons
between as197 and x12; it uses some stock series, some simulated series and
series from Oleh's dataset. Have fun.
I'm having fun already! Thanks.
Allin
7:55 p.m.
Dear Riccardo,
[1,4]: 0.72 vs 3.96; as197 probably faulty (d loglik = 8.34761)
How to see specification?
In the presence of a unit root iterations should bump
into the unit circle
BUT the theorem on optimal properties of MLE
requires data should be transformed to random
sample
So, in the presence of a unit root the final
value of log likelihood is not very meaningful
The main thing here - bumping the unit circle border
In this case a user can easily infer that model
is misspecified
The difference in final values of loglik is meaningful
only if there are no unit roots
Oleh
20 березня 2018, 15:28:23, від "Riccardo (Jack) Lucchetti"
<r.lucchetti(a)univpm.it>:
On Tue, 20 Mar 2018, > oleg_komashko(a)ukr.net wrote:
> Dear all,
> It seems the list now
> is more powerful with attachments
> so 792 M4 economic monthly series
> as .gdt file attached
I'm attaching a script file that performs some performance comparisons
between as197 and x12; it uses some stock series, some simulated series
and series from Oleh's dataset. Have fun.
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti@univpm.ithttp://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
_______________________________________________
Gretl-devel mailing list
Gretl-devel@lists.wfu.eduhttp://lists.wfu.edu/mailman/listinfo/gretl-devel
2229
days inactive
2229
days old
3 comments
3 participants
participants (3)
-
Allin Cottrell -
oleg_komashko＠ukr.net -
Riccardo (Jack) Lucchetti