gretl speed + windows
by Zhukov Pavel
Why gretl running on windows is extremely slow?
same script running on linux takes 20 sec
but in widows over 2 min.
if script running on linux take 2-3 min then in windows it take over 10 min?
18 years, 4 months
alert: database corrupted!
by Allin Cottrell
The gretl package comes with one US macroeconomic database,
conatining data from the Federal Reserve Bank of St Louis.
I'm afraid that the version of this database that was distributed
with gretl 1.5.1 may be corrupted, in that some of the series
lengths are wrong, so the series as advertised in the database
index are out of phase with the true data series.
My apologies for this oversight. You can fix this, if you have
Internet access, as follows:
* Go to the File menu
* Select Browse databases, on database server
* Scroll down and select the "fedstl" database
* Click the "Install" button
This will install a corrected (and updated) version of the St
Louis Fed database.
--
Allin Cottrell
Department of Economics
Wake Forest University, NC
18 years, 4 months
series element
by Zhukov Pavel
morning.
How i can replace some element in a vector (series)?
e.g.
series n=0;
series n(1)=1
^ this is not working! :(
18 years, 4 months
gretl development news
by Allin Cottrell
Hello all,
This is a very busy time of the academic year for me, so I've had
to cut back a bit on the time I spend on gretl. Nonetheless, there
are some nice things in the pipeline and I thought I'd let people
know about some of them.
1) Up till now, we've had conditional ML estimation of ARIMA models
in "native" gretl code, but exact ML has been farmed out to
x-12-arima and TRAMO/SEATS. Well, now we have native exact ML using
the Kalman filter. The Kalman filter is now accessible via the
C-language API of libgretl. I haven't yet attempted to make it
accessible via script commands: the interface will necessarily be
quite complicated. The Kalman/ARIMA functionality is available in
gretl CVS and the current Windows snapshot
http://ricardo.ecn.wfu.edu/pub/gretl/gretl_install.exe
2) There's a new (and, I think, greatly improved) mechanism for
saving gretl GUI sessions. We now save sessions (something like
OpenOffice) as a zipfile containing the graph files from the
session, the session models represented in XML, and any user-defined
matrices and functions, also in XML. We also save the state of the
"model table" and the "graph page". In my testing so far this seems
much more reliable than the old method. One thing I still have to
work on is a conversion mechanism, for backward compatibility, so
that gretl will be able to salvage something from session files
saved in the old format.
3) User-defined functions. I think these are an important element
in the future of gretl, since they allow people who don't code in C,
but who know gretl, to contribute functionality. I'm working on an
XML representation of functions, so we can have specialized
functions with their own "help" documentation, author info, and so
on. This is not finished yet; I'll say more when it's nearer to
being ready.
Allin.
--
Allin Cottrell
Department of Economics
Wake Forest University, NC
18 years, 4 months
Re: Gretl-users Digest, Vol 24, Issue 3
by John Paravantis
Again, X-12-ARIMA somehow manages to estimate ALL ARIMA models,
including the ones that the default routine CANNOT.
John
gretl-users-request(a)ricardo.ecn.wfu.edu wrote:
>> Coming now to the crunch of the matter, there appear to exist QUITE A
>> FEW ARIMA models that CANNOT be estimated with gretl which gives a
>> dialog box stating "The convergence criterion was not met".
>> UNFORTUNATELY, there is NO OPTION to increase the number of iterations
>> (or make the convergence threshold more lax). These models include:
>>
>> ARIMA(1,2,1)
>> ARIMA(0,2,1)
>>
>> It should be noted that these models MAY BE ESTIMATED with X-12_ARIMA
>> but not with the "default" ARIMA routine.
>>
>>
> I do not know what models yo got estimated, I only know that the
> models you
> sent in your previous message was not working. I suppose you may have success
> with models such as ARMA(1,2,0) or ARMA(2,2,0), this models are linear, so
> maximum likelihood has an analitical solution and it is equivalent to OLS. So
> in this case you obtain a direct estimation. The models with a MA part (and
> the ARxARs as well) are not linear and finding the maximum of the likelihood
> function require an iterative optimization routine. The difference between
> the estimated likelihood from one iteration to the following one may be large
> if you have few data, and so the process is difficult to converge.
>
>
18 years, 5 months
Re: Gretl-users Digest, Vol 24, Issue 2
by John Paravantis PhD
Some points on the feedback appearing below:
- SPSS, Minitab etc all successfully estimate these ARIMA models and it
is left up to the user to decide how much trust should be placed with
the estimates. Gretl fails to provide an estimate and I find this annoying.
- X-12-ARIMA also manages to provide an estimate in ALL cases.
- If the data are NOT enough, then how come SOME ARIMA specifications
ARE estimated?
I also checked the message at
http://ricardo.ecn.wfu.edu/pipermail/gretl-users/2006-February/000499.html
How about making the number of iterations and tolerance available in the
dialog box version of ARIMA?
Many thanks for your time.
John
>> Now, on the bleaker side, I seem to face some problems with the routine.
>> First of all you may download my test file from
>>
>> http://paravantis.com/cars100.gdt
>>
>> (It represents the number of cars per 100 people in Greece from 1970 to
>> 2003.)
>>
>
> I think you have few observations to work with standard time series tecniques.
> In ARIMA models you do not have unbiadness in the estimators, you only have
> consistency, and this is a property that justifies good estimations only with
> a large number of observations. The sample correlograms are only consistent
> as well, so you may not trust in them with so few observations.
>
> Few observations produces large standard deviations of the coefficients so the
> estimations have a large variance (they are very inaccurate) and all the
> tests tend to be not rejected.
>
>
>> Coming now to the crunch of the matter, there appear to exist QUITE A
>> FEW ARIMA models that CANNOT be estimated with gretl which gives a
>> dialog box stating "The convergence criterion was not met".
>> UNFORTUNATELY, there is NO OPTION to increase the number of iterations
>> (or make the convergence threshold more lax). These models include:
>>
>> ARIMA(1,2,1)
>> ARIMA(0,2,1)
>>
>> It should be noted that these models MAY BE ESTIMATED with X-12_ARIMA
>> but not with the "default" ARIMA routine.
>>
>
> I think the problem here is that you have so few observations and the variance
> of the coefficient estimator is so large that the change in the estimations
> from one iteration to the following is so large that gretl cannot converge.
>
> I asked in a previous message to this list to have a method for changing the
> number of iterations. You may read the response in
>
> http://ricardo.ecn.wfu.edu/pipermail/gretl-users/2006-February/000499.html
>
> but I think probably this will not work in your case.
>
>
>> Here is what I would appreciate having as feedback from the respected
>> gretl community:
>>
>> 1. How do you stand on the differencing issue? Do you side with 1st or
>> 2nd differences? Does it really matter since we can "correct" by
>> including more AR or MA terms?
>>
>
> Apart form the graphics (and with your graphic I would vote for I(2)), we
> normally use a test (as ADF: in /variable/augmented Dickey-Fuller test) but
> with your data you will see that all the tests are not rejected, and this is
> only because your data have no enough information.
>
>
>> 2. What causes the problem with the default ARIMA routine?
>>
>
> Too few observations.
>
>
>> Is it "safe"
>> to use X-12-ARIMA in all cases (even with unseasonal data)?
>>
>
> No. In some cases you may obtain an estimation, but in your case you may not
> trust in it because you have a very large variance.
>
>
>
18 years, 5 months
ARIMA issues
by John Paravantis PhD
First of all, kudos to Allin and the development team for making an
ARIMA (as opposed to ARMA) routine available in the latest version of
gretl! I personally find this of immense utility in my classes as
students do not have to back calculate raw level forecasts from
differenced data.
Now, on the bleaker side, I seem to face some problems with the routine.
First of all you may download my test file from
http://paravantis.com/cars100.gdt
(It represents the number of cars per 100 people in Greece from 1970 to
2003.)
The data set includes 3 columns: Cars100 (cars per 100 people), DIFF1_Ca
(1st differences) and DIFF2_Ca (2nd differences). Now, in the past Allin
indicated to me that he felt that taking the 2nd differences of Cars100
was probably OVERDIFFERENCING. Yet, using correlograms and taking into
account which of these series has the SMALLEST STANDARD DEVIATION, I
think that the second differences are "best" (and it is kept in mind
that mild underdifferencing may be corrected with more AuroRegressive
terms while mild overdifferencing may be corrected with more Moving
Average terms).
Coming now to the crunch of the matter, there appear to exist QUITE A
FEW ARIMA models that CANNOT be estimated with gretl which gives a
dialog box stating "The convergence criterion was not met".
UNFORTUNATELY, there is NO OPTION to increase the number of iterations
(or make the convergence threshold more lax). These models include:
ARIMA(1,2,1)
ARIMA(0,2,1)
It should be noted that these models MAY BE ESTIMATED with X-12_ARIMA
but not with the "default" ARIMA routine.
Here is what I would appreciate having as feedback from the respected
gretl community:
1. How do you stand on the differencing issue? Do you side with 1st or
2nd differences? Does it really matter since we can "correct" by
including more AR or MA terms?
2. What causes the problem with the default ARIMA routine? Is it "safe"
to use X-12-ARIMA in all cases (even with unseasonal data)?
Thanking you all for you time,
John Paravantis
University of Piraeus
Greeece
18 years, 5 months
log-likelihood
by Ignacio Díaz-Emparanza
First of all I want to report that I installed new version gretl-1.5.1 in
mandriva-linux by means of the gretl-1.5.1-1gtk2.i586.rpm and all is working
perfectly.
But now, ... one step ahead:
I am trying to program maximum likelihood in the "prediction error
decomposition form" (see Harvey 1990 "Time Series Models", p.91)
but I have a problem. I need to run the kfilter script from inside a
"mle" (maximum likelihood estimation) group, and it is not possible.
¿Any suggestion?
(see scripts in http://www.bl.ehu.es/~etpdihei/gretl/kloglike.inp
http://www.bl.ehu.es/~etpdihei/gretl/kfilterft.inp )
include kfilterft.inp
mle logl = -log(ft)-st
# Next line is the function call, here is the problem
(filtered, ft) = kfilter Y, a0, p0, bT, Z, Sigma_w, sigma_e
series vt = Y-filtered
series st = (vt**2)/(ft*T)
param sigma_e
end mle
--
Ignacio Díaz-Emparanza
Dpto. de Economía Aplicada III (Econometría y Estadística)
UPV-EHU
18 years, 5 months
