12:29 p.m.
Attached you will find a script that applies a Hodrick-Prescott filter by
means of a Kalman filter. I hope in future versions of gretl the "function"
command can accept matrices as parameters, so this will do easier to apply
the KF. We could advance then in programming, for example, Maximum Likelihood
trough "prediction error decomposition" and then Structural Time Series
Models.
--
Ignacio Díaz-Emparanza
Dpto. de Economía Aplicada III (Econometría y Estadística)
UPV-EHU
12:43 p.m.
New subject: Close button vol.2
Also "X" - close button is not working when you want to configure Robust
standard errors.
_________________________________________________________________
Be the first to hear what's new at MSN - sign up to our free newsletters!
http://www.msn.co.uk/newsletters
3:25 p.m.
New subject: Close button vol.2
On Thu, 16 Mar 2006, john w wrote:
Also "X" - close button is not working when you want to
configure Robust
standard errors.
OK, I think I see what you mean: "X" is the destroy button added by
Windows to the outer frame of the dialog box? Yes, this should be
equivalent to selecting Cancel.
Allin Cottrell
8:49 a.m.
New subject: Close button vol.2
Yes, that's wright! It's the kill button. For those three fatures it's not
working well.
From: Allin Cottrell <cottrell(a)wfu.edu>
Reply-To: Gretl list <gretl-users(a)ricardo.ecn.wfu.edu>
To: Gretl list <gretl-users(a)ricardo.ecn.wfu.edu>
Subject: Re: [Gretl-users] Close button vol.2
Date: Thu, 16 Mar 2006 10:25:27 -0500 (EST)
On Thu, 16 Mar 2006, john w wrote:
>Also "X" - close button is not working when you want to configure Robust
>standard errors.
OK, I think I see what you mean: "X" is the destroy button added by Windows
to the outer frame of the dialog box? Yes, this should be equivalent to
selecting Cancel.
Allin Cottrell
_______________________________________________
Gretl-users mailing list
Gretl-users(a)ricardo.ecn.wfu.edu
http://ricardo.ecn.wfu.edu/mailman/listinfo/gretl-users
_________________________________________________________________
The new MSN Search Toolbar now includes Desktop search!
http://toolbar.msn.co.uk/
5:09 p.m.
On Thu, March 16, 2006 13:29, Ignacio Díaz-Emparanza wrote:
Attached you will find a script that applies a Hodrick-Prescott
filter by
means of a Kalman filter. I hope in future versions of gretl the "function"
command can accept matrices as parameters, so this will do easier to apply
the KF. We could advance then in programming, for example, Maximum Likelihood
trough "prediction error decomposition" and then Structural Time Series
Models.
With current CVS you already can have matrices as arguments. What you can't do
(and I admit that needs to be addressed) is *return* matrices from functions.
The following is an example function, basically ripped from your script with
some optimisations, which does the forward pass and returns the state as a
series.
function kfilt(series y, matrix a0, matrix p0, matrix bT, matrix Z, \
matrix bT, matrix Sigma_w, scalar sigma_e)
# Forward solution
#
scalar T = rows(Z)
scalar m = rows(a0)
matrix at_t = a0
matrix pt_t = p0
matrix at = zeros(m,T)
printf "\nStarting the filter process\n"
printf "\n...Filtering...\n"
loop for i=1..T
# Prediction equations
# eq. (2.2a)
matrix at_t = bT*at_t
# eq. (2.2b)
matrix pt_t1 = bT*pt_t*transp(bT)+Sigma_w
# eq. (2.3c)
matrix zt = Z[$i,]
matrix H = pt_t1*zt'
matrix f = zt*H + sigma_e
if i>1
matrix pstar_t=pt_t*transp(bT)*inv(pt_t1)
endif
#Updating equations
# eq. (2.4a)
scalar e = y[$i] - zt*at_t
matrix at_t = at_t + H*(e/f)
# eq (2.4b)
matrix pt_t = pt_t1 - H*H' * (1/f)
matrix at[,$i]=at_t
if i==1
matrix pt=pt_t
else
if i==2
matrix pt=pt~pt_t
matrix pstar=pstar_t
else
matrix pt=pt~pt_t
matrix pstar=pstar~pstar_t
endif
endif
end loop
series filtered = at[1,]
printf "\nFilter done\n"
return series filtered
end function
Riccardo "Jack" Lucchetti
Dipartimento di Economia
Facoltà di Economia "G. Fuà"
Ancona
6634
days inactive
6635
days old
5 comments
4 participants
participants (4)
-
Allin Cottrell -
Ignacio Díaz-Emparanza -
john w -
Riccardo Jack Lucchetti