Bug
by Ignacio Díaz-Emparanza
Bug report:
Running Gretl 1.2.3 (in Spanish, installed from the "generic" rpm) on Linux
Mandrake 9.2 (KDE).
I open the Ramanathan's data set data3-6.gdt and ask for an OLS estimation,
then I ask for "Tests-Autocorrelation" and put a number for the lags into the
dialog box, the result is
gretl error: invalid option '-m'
--
Ignacio Díaz-Emparanza
Dpto. de Economía Aplicada III (Econometría y Estadística)
UPV-EHU
http://www.bl.ehu.es/~etpdihei/
20 years, 7 months
wishes
by Ignacio Díaz-Emparanza
Allin (and friends),
for a person like me, who works mainly with time series the native code for
the ARMA estimation is very wellcome, but would it be very difficult to do it
slightly more general? In particular I would like to have:
- Multiplicative seasonal models ( you can use a dialog box similar to the one
in the tramo/seats plugin)
- External regressors (explanatory variables) added to the ARIMA equation. (So
we may consider the estimation of intervention models).
Thanks in advance,
--
Ignacio Díaz-Emparanza
Dpto. de Economía Aplicada III (Econometría y Estadística)
UPV-EHU
http://www.bl.ehu.es/~etpdihei/
20 years, 7 months
Translation of gretl documentation
by Cristian Rigamonti
Hi Allin and friends, having completed the Italian translation of gretl,
I've started thinking about translating some of its documentation, and I
have a couple of questions (mainly to Allin, but maybe other users will
come up with useful suggestions):
- I'd start from the web page: can I just download the HTML file and
translate it, or is there another source document (SGML/XML...) for
it? Is it stored somewhere in the CVS (I've checked, but it doesn't
seem...)?
- I'd then go on with the online help files; I've noticed that some
content from the /share/*.hlp files overlaps with the content of the
/doc/chapters/cmdref.xml file (the Manual's "Command Reference"
chapter). I had a look at their CVS logs and noticed that the update
timestamps are quite regular, so I was wondering if the help files
were somehow generated automatically from the xml file (in that case
it would be more efficient to translate the latter).
More generally, I'd like to know something about the future of gretl
documentation, e.g. are there any parts of the documentation that will
be heavily modified, so that it's not worth to start translating them
now? Are there any plans to switch to other formats, or something? I
would be glad to help, for what I can do.
Thanks for reading.
Cri
--
[ Cristian Rigamonti - cri(a)linux.it - +39 347 1043466 ]
[ GPG/PGP Key-Id: 943A5F0E - http://www.linux.it/~cri/cri.asc ]
20 years, 7 months
NIST tests
by Ignacio Díaz-Emparanza
Allin (and friends),
I have been taking a look at the new menu entry "NIST test suite", and I was
surprised by the results. I was looking at them with the spanish gretl 1.2.2
version using the comma as the decimal separator. The summary result was
Summary of NIST linear regression test results:
* number of tests carried out: 11
* reference data files missing or corrupted: 0
* unexpected errors in estimation of models: 1
* poor or unacceptable results with libgretl: 10
* cases where results from the gretl GMP plugin disagreed with the NIST
certified values, at 12 significant figures: 9
I was really surprised as I remembered that the manual reported some good
results, so I looked at the web page and compared my results with the results
shown there noticing that they are also completely different. I have run the
same tests setting the decimal point as decimal separator, and now the
results are quite good. So I there is an error when the NIST tests are run
with the decimal comma.
Enclosed you will find three files: nist-c.out has the "basic results" of the
NIST tests when I use a comma as decimal separator, nist-p.out has the same
with the decimal point and nist-ext.out has the extended results when I use
the comma again. I hope this will help you to detect the error.
Tell me if you need any more.
Best regards,
--
Ignacio Díaz-Emparanza
Dpto. de Economía Aplicada III (Econometría y Estadística)
UPV-EHU
http://www.bl.ehu.es/~etpdihei/
20 years, 8 months