1:59 p.m.
Hello all,
We recently redesigned the presentation of coefficients, standard
errors, and so on, in the printing of model results in gretl.
We're now thinking of revamping the presentation of the additional
results that follow the table of coefficients (i.e., sum of
squared residuals, R^2, loglikelihood, etc.).
This has been prompted by Christopher Dougherty, who intends to
use gretl as the standard software in a new edition of his
Introduction to Econometrics (Oxford). He'll be printing a bunch
of regression results and he asked if we'd consider using a more
compact format, along the lines of Eviews.
It's true, our current presentation uses a lot of vertical space.
I've mocked up an alternative compact presentation and I'd like to
hear what people think of this. I'm particularly keen to hear
from translators, because it could be a problem translating the
abbreviated strings.
Please see the example at
http://ricardo.ecn.wfu.edu/~cottrell/dougherty/model.html
where I show a given model as displayed by current gretl, the
proposed modification, and Stata and Eviews output for comparison.
Note that for a translation of the compact table to work
correctly, all the abbreviated strings (such as "Mean dependent
var", "Sum squared resid") should be no greater than 19 characters
in length. Does this look doable?
If translation is not feasible, I'm thinking of including the
compact presentation as an English-only option -- but obviously it
would be preferable to maintain consistency across languages if
possible.
Allin.
2:34 p.m.
Yep, it looks doable for the Turkish translation and quite impressive
I must say.
Cheers
Talha
On Thu, Nov 20, 2008 at 3:59 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
Hello all,
We recently redesigned the presentation of coefficients, standard
errors, and so on, in the printing of model results in gretl.
We're now thinking of revamping the presentation of the additional
results that follow the table of coefficients (i.e., sum of
squared residuals, R^2, loglikelihood, etc.).
This has been prompted by Christopher Dougherty, who intends to
use gretl as the standard software in a new edition of his
Introduction to Econometrics (Oxford). He'll be printing a bunch
of regression results and he asked if we'd consider using a more
compact format, along the lines of Eviews.
It's true, our current presentation uses a lot of vertical space.
I've mocked up an alternative compact presentation and I'd like to
hear what people think of this. I'm particularly keen to hear
from translators, because it could be a problem translating the
abbreviated strings.
Please see the example at
http://ricardo.ecn.wfu.edu/~cottrell/dougherty/model.html
where I show a given model as displayed by current gretl, the
proposed modification, and Stata and Eviews output for comparison.
Note that for a translation of the compact table to work
correctly, all the abbreviated strings (such as "Mean dependent
var", "Sum squared resid") should be no greater than 19 characters
in length. Does this look doable?
If translation is not feasible, I'm thinking of including the
compact presentation as an English-only option -- but obviously it
would be preferable to maintain consistency across languages if
possible.
Allin.
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
"Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment." - Benjamin Franklin (1706-1790)
--
2:42 p.m.
El Thursday 20 November 2008 14:59:29 Allin Cottrell escribió:
Hello all,
...
Please see the example at
http://ricardo.ecn.wfu.edu/~cottrell/dougherty/model.html
where I show a given model as displayed by current gretl, the
proposed modification, and Stata and Eviews output for comparison.
Note that for a translation of the compact table to work
correctly, all the abbreviated strings (such as "Mean dependent
var", "Sum squared resid") should be no greater than 19 characters
in length. Does this look doable?
If translation is not feasible, I'm thinking of including the
compact presentation as an English-only option -- but obviously it
would be preferable to maintain consistency across languages if
possible.
Allin.
I like it, and I don't see much problem for Spanish.
--
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.et.bs.ehu.es
3:40 p.m.
Looks good, and of course is doable.
The only thing is that condensed stuff like "sum squared resid" has been
known in English for a long time, so everybody knows what it is. While
in other languages I guess we translators just make up expressions that
occur to us and it is much less clear to the general public how to
expand such abbreviations. But this is just the implication of gretl's
localization, which of course is a good thing; so we boldly go where no
econometrician has gone before...
cheers,
sven
Am 20.11.2008 14:59, Allin Cottrell schrieb:
Hello all,
We recently redesigned the presentation of coefficients, standard
errors, and so on, in the printing of model results in gretl.
We're now thinking of revamping the presentation of the additional
results that follow the table of coefficients (i.e., sum of
squared residuals, R^2, loglikelihood, etc.).
This has been prompted by Christopher Dougherty, who intends to
use gretl as the standard software in a new edition of his
Introduction to Econometrics (Oxford). He'll be printing a bunch
of regression results and he asked if we'd consider using a more
compact format, along the lines of Eviews.
It's true, our current presentation uses a lot of vertical space.
I've mocked up an alternative compact presentation and I'd like to
hear what people think of this. I'm particularly keen to hear
from translators, because it could be a problem translating the
abbreviated strings.
Please see the example at
http://ricardo.ecn.wfu.edu/~cottrell/dougherty/model.html
where I show a given model as displayed by current gretl, the
proposed modification, and Stata and Eviews output for comparison.
Note that for a translation of the compact table to work
correctly, all the abbreviated strings (such as "Mean dependent
var", "Sum squared resid") should be no greater than 19 characters
in length. Does this look doable?
If translation is not feasible, I'm thinking of including the
compact presentation as an English-only option -- but obviously it
would be preferable to maintain consistency across languages if
possible.
Allin.
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
4:03 p.m.
I think the proposed presentation of model results in English looks great,
but in (Brazilian) Portuguese I think it doesn´t look good (I'd attached a
.txt file with a example). When we translate scientific texts (or
expressions) from English to Portuguese we usually need more words (and
space) to keep the original meaning of the expression. The main reason was
pointed by Sven:
"The only thing is that condensed stuff like 'sum squared resid' has been
known in English for a long time, so everybody knows what it is. While
in other languages I guess we translators just make up expressions that
occur to us and it is much less clear to the general public how to
expand such abbreviations".
Regards,
Henrique C. de Andrade
Ph.D. Program in Applied Economics
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
4:15 p.m.
As I mentioned before I am perfectly OK for the Turkish translation
but here is a suggestion that could be useful for other languages:
Now, everybody knows that S.E. is for standard error but suppose this
would be O.H. for Turkish which may or may not be so obvious.
Consequently, could there be an additional line or two reserved at the
bottom saying "O.H. stands for olcunlu hata" or something like that?
This can be optional and used for languages such as Portuguese while
other leave that field blank. What do you think?
Talha
2008/11/20 Henrique <henrique.coelho(a)gmail.com>:
I think the proposed presentation of model results in English looks
great,
but in (Brazilian) Portuguese I think it doesn´t look good (I'd attached a
.txt file with a example). When we translate scientific texts (or
expressions) from English to Portuguese we usually need more words (and
space) to keep the original meaning of the expression. The main reason was
pointed by Sven:
"The only thing is that condensed stuff like 'sum squared resid' has been
known in English for a long time, so everybody knows what it is. While
in other languages I guess we translators just make up expressions that
occur to us and it is much less clear to the general public how to
expand such abbreviations".
Regards,
Henrique C. de Andrade
Ph.D. Program in Applied Economics
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
"Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment." - Benjamin Franklin (1706-1790)
--
4:23 p.m.
Now, everybody knows that S.E. is for standard error but suppose
this
would be O.H. for Turkish which may or may not be so obvious.
Consequently, could there be an additional line or two reserved at the
bottom saying "O.H. stands for olcunlu hata" or something like that?
This can be optional and used for languages such as Portuguese while
other leave that field blank. What do you think?
Talha
I think this is a great idea!
--
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
4:44 p.m.
On Thu, 20 Nov 2008, Talha Yalta wrote:
Now, everybody knows that S.E. is for standard error but suppose
this
would be O.H. for Turkish which may or may not be so obvious.
Consequently, could there be an additional line or two reserved at the
bottom saying "O.H. stands for olcunlu hata" or something like that?
I could add a translatable string -- to appear under the table --
that says something like "NOTES ON MODEL STATISTICS ABBREVIATIONS
GO HERE", and arrange for it to be printed only if the current
language is not English.
Allin.
4:53 p.m.
I could add a translatable string -- to appear under the table --
that says something like "NOTES ON MODEL STATISTICS ABBREVIATIONS
GO HERE", and arrange for it to be printed only if the current
language is not English.
Does that mean all languages except English have to use
it? Well,
eitherway sounds good to me. Infact, this could be useful for making
the abbreviations more obvious, helping adoption by others.
Allin.
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
"Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment." - Benjamin Franklin (1706-1790)
--
5:31 p.m.
On Thu, 20 Nov 2008, Talha Yalta wrote:
> I could add a translatable string -- to appear under the table
--
> that says something like "NOTES ON MODEL STATISTICS ABBREVIATIONS
> GO HERE", and arrange for it to be printed only if the current
> language is not English.
Does that mean all languages except English have to use it? Well,
eitherway sounds good to me. Infact, this could be useful for making
the abbreviations more obvious, helping adoption by others.
You'd have the option. I'm adding to CVS a string,
"note on model statistics abbreviations here"
The criterion for printing it is: print it only if the version
retrieved by gettext is not the same as the English. Thus it'll
never be printed if gretl is running in English, and neither will
it be printed when gretl is running in another language if it's
not translated.
So if you'd like to make use of it, "translate" it -- that is,
replace it with something useful -- otherwise leave it
untranslated and it won't be seen.
Allin.
6:55 p.m.
You'd have the option. I'm adding to CVS a string,
"note on model statistics abbreviations here"
Sounds great to me. Cheers
Talha
--
"Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment." - Benjamin Franklin (1706-1790)
--
3:55 p.m.
On Thu, 20 Nov 2008, Allin Cottrell wrote:
Please see the example at
http://ricardo.ecn.wfu.edu/~cottrell/dougherty/model.html
where I show a given model as displayed by current gretl, the
proposed modification, and Stata and Eviews output for comparison.
I quite like it. Attached you'll find a patch against CVS that does 3
things:
1) toggles compaction on an environment variable, to make testing easier
2) suppresses DW for non-time-series data
3) organises stuff in a more logical (IMHO) way than eviews. Example:
Model 1: OLS estimates using the 14 observations 1-14
Dependent variable: price
coefficient std. error t-ratio p-value
--------------------------------------------------------
const 52.3509 37.2855 1.404 0.1857
sqft 0.138750 0.0187329 7.407 8.20E-06 ***
Mean dependent var 317.4929 S.D. dependent var 88.49816
Sample size 14 Sum squared resid 18273.57
R-squared 0.820522 Adjusted R-squared 0.805565
S.E. of regression 39.02304 Log-likelihood -70.08421
Akaike criterion 144.1684 Schwarz criterion 145.4465
F(1, 12) 54.86051 P-value(F) 8.20e-06
Comments welcome!
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucchetti(a)univpm.it
http://www.econ.univpm.it/lucchetti
4:06 p.m.
2) suppresses DW for non-time-series data
In addition, I wonder
whether gretl checks all the assumptions of DW
before diplaying it:
a)- The constant exists,
b)- No lagged dependent variable as explanatory variable,
c)- No missing observations.
Cheers
Talha
On Thu, Nov 20, 2008 at 5:55 PM, Riccardo (Jack) Lucchetti
<r.lucchetti(a)univpm.it> wrote:
> On Thu, 20 Nov 2008, Allin Cottrell wrote:
>
>
>> Please see the example at
>> http://ricardo.ecn.wfu.edu/~cottrell/dougherty/model.html
>> where I show a given model as displayed by current gretl, the
>> proposed modification, and Stata and Eviews output for comparison.
>
> I quite like it. Attached you'll find a patch against CVS that does 3
> things:
>
> 1) toggles compaction on an environment variable, to make testing easier
2) suppresses DW for non-time-series data
> 3) organises
stuff in a more logical (IMHO) way than eviews. Example:
>
> Model 1: OLS estimates using the 14 observations 1-14
> Dependent variable: price
>
> coefficient std. error t-ratio p-value
> --------------------------------------------------------
> const 52.3509 37.2855 1.404 0.1857
> sqft 0.138750 0.0187329 7.407 8.20E-06 ***
>
> Mean dependent var 317.4929 S.D. dependent var 88.49816
> Sample size 14 Sum squared resid 18273.57
> R-squared 0.820522 Adjusted R-squared 0.805565
> S.E. of regression 39.02304 Log-likelihood -70.08421
> Akaike criterion 144.1684 Schwarz criterion 145.4465
> F(1, 12) 54.86051 P-value(F) 8.20e-06
>
> Comments welcome!
>
>
>
> Riccardo (Jack) Lucchetti
> Dipartimento di Economia
> Università Politecnica delle Marche
>
> r.lucchetti(a)univpm.it
> http://www.econ.univpm.it/lucchetti
> _______________________________________________
> Gretl-devel mailing list
> Gretl-devel(a)lists.wfu.edu
> http://lists.wfu.edu/mailman/listinfo/gretl-devel
>
--
"Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment." - Benjamin Franklin (1706-1790)
--
4:08 p.m.
On Thu, 20 Nov 2008, Riccardo (Jack) Lucchetti wrote:
I quite like it. Attached you'll find a patch against CVS that
does 3 things:
1) toggles compaction on an environment variable, to make testing easier
2) suppresses DW for non-time-series data
3) organises stuff in a more logical (IMHO) way than eviews. Example:
...
Oh, another thing. I skipped the Hannan-Quinn criterion for consistency
with Allin's original proposal, but I'd really prefer to keep it.
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucchetti(a)univpm.it
http://www.econ.univpm.it/lucchetti
4:46 p.m.
On Thu, 20 Nov 2008, Riccardo (Jack) Lucchetti wrote:
On Thu, 20 Nov 2008, Riccardo (Jack) Lucchetti wrote:
> I quite like it. Attached you'll find a patch against CVS that
> does 3 things:
>
> 1) toggles compaction on an environment variable, to make testing easier
> 2) suppresses DW for non-time-series data
> 3) organises stuff in a more logical (IMHO) way than eviews. Example:
This looks fine to me.
Oh, another thing. I skipped the Hannan-Quinn criterion for
consistency with Allin's original proposal, but I'd really
prefer to keep it.
I'd be happy to keep it, I was just trying to come up with an even
number of basic stats.
Allin.
10:47 p.m.
On Thu, 20 Nov 2008, Allin Cottrell wrote:
> Oh, another thing. I skipped the Hannan-Quinn criterion for
> consistency with Allin's original proposal, but I'd really
> prefer to keep it.
I'd be happy to keep it, I was just trying to come up with an even
number of basic stats.
Would this be too much? One nice thing about it is that it reminds
me of archaic versions of gretl :-)
Index: lib/src/modelprint.c
===================================================================
RCS file: /cvsroot/gretl/gretl/lib/src/modelprint.c,v
retrieving revision 1.356
diff -u -r1.356 modelprint.c
--- lib/src/modelprint.c 20 Nov 2008 18:01:39 -0000 1.356
+++ lib/src/modelprint.c 20 Nov 2008 22:42:46 -0000
@@ -2568,9 +2568,6 @@
pprintf(prn, "%-20s%s %-20s%s\n", _("S.E. of regression"),
print_eight(x1, pmod->sigma),
_("Log-likelihood"), print_eight(x2, pmod->lnL));
- pprintf(prn, "%-20s%s %-20s%s\n", _("Akaike criterion"),
- print_eight(x1, pmod->criterion[C_AIC]),
- _("Schwarz criterion"), print_eight(x2,
pmod->criterion[C_BIC]));
pprintf(prn, "%-20s%s %-20s%s\n",
fstr, print_eight(x1, pmod->fstt),
_("P-value(F)"), print_eight(x2, pvF));
@@ -2580,6 +2577,13 @@
print_eight(x2, pmod->dw));
}
pputc(prn, '\n');
+ pputs(prn, _("Information criteria"));
+ pputs(prn, ":\n");
+ pprintf(prn, "AIC: %14.5f BIC: %14.5f HQC: %14.5f\n",
+ pmod->criterion[C_AIC],
+ pmod->criterion[C_BIC],
+ pmod->criterion[C_HQC]);
+ pputc(prn, '\n');
if (strcmp(note, _(note))) {
pputs(prn, _(note));
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucchetti(a)univpm.it
http://www.econ.univpm.it/lucchetti
12:48 p.m.
Apart from the looks, I'm not sure why Akaike should be preferred over
Hannan-Quinn in terms of econometrics. I'm sure AIC is good for some
objectives, but AFAIK it is not consistent in the sense that it doesn't
pick the true model even asymptotically, in contrast to Hannan-Quinn.
Am I mixing things up here, or are these thoughts irrelevnant for the
issue we're discussing right now? Otherwise I would tend to either keep
all or ditch AIC.
cheers,
sven
Riccardo (Jack) Lucchetti schrieb:
On Thu, 20 Nov 2008, Allin Cottrell wrote:
>> Oh, another thing. I skipped the Hannan-Quinn criterion for
>> consistency with Allin's original proposal, but I'd really
>> prefer to keep it.
>
> I'd be happy to keep it, I was just trying to come up with an even
> number of basic stats.
Would this be too much? One nice thing about it is that it reminds me of
archaic versions of gretl :-)
Index: lib/src/modelprint.c
===================================================================
RCS file: /cvsroot/gretl/gretl/lib/src/modelprint.c,v
retrieving revision 1.356
diff -u -r1.356 modelprint.c
--- lib/src/modelprint.c 20 Nov 2008 18:01:39 -0000 1.356
+++ lib/src/modelprint.c 20 Nov 2008 22:42:46 -0000
@@ -2568,9 +2568,6 @@
pprintf(prn, "%-20s%s %-20s%s\n", _("S.E. of regression"),
print_eight(x1, pmod->sigma),
_("Log-likelihood"), print_eight(x2, pmod->lnL));
- pprintf(prn, "%-20s%s %-20s%s\n", _("Akaike criterion"),
- print_eight(x1, pmod->criterion[C_AIC]),
- _("Schwarz criterion"), print_eight(x2,
pmod->criterion[C_BIC]));
pprintf(prn, "%-20s%s %-20s%s\n",
fstr, print_eight(x1, pmod->fstt),
_("P-value(F)"), print_eight(x2, pvF));
@@ -2580,6 +2577,13 @@
print_eight(x2, pmod->dw));
}
pputc(prn, '\n');
+ pputs(prn, _("Information criteria"));
+ pputs(prn, ":\n");
+ pprintf(prn, "AIC: %14.5f BIC: %14.5f HQC: %14.5f\n",
+ pmod->criterion[C_AIC],
+ pmod->criterion[C_BIC],
+ pmod->criterion[C_HQC]);
+ pputc(prn, '\n');
if (strcmp(note, _(note))) {
pputs(prn, _(note));
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucchetti(a)univpm.it
http://www.econ.univpm.it/lucchetti
------------------------------------------------------------------------
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
2 p.m.
On Fri, 21 Nov 2008, Sven Schreiber wrote:
Apart from the looks, I'm not sure why Akaike should be
preferred over Hannan-Quinn in terms of econometrics. I'm sure
AIC is good for some objectives, but AFAIK it is not consistent
in the sense that it doesn't pick the true model even
asymptotically, in contrast to Hannan-Quinn.
Am I mixing things up here, or are these thoughts irrelevnant
for the issue we're discussing right now? Otherwise I would tend
to either keep all or ditch AIC.
I agree, the AIC is not all that great. But it's the best known
of the information criteria and if we dropped it I guess we would
have people saying, why don't you print the AIC?
Here's another go at this topic. I've put up
http://ricardo.ecn.wfu.edu/~cottrell/dougherty/table.html
showing a few variants for folks to consider.
* Version A is what I posted yesterday. It is closely (too
closely?) based on eviews and does not have a space for the HQC;
also it prints the Durbin-Watson statistic regardless of whether
or not the data are time series (but does not print rho). We
probably don't want this.
* Version B is what Jack suggested yesterday (although it doesn't
fully correspond with his preferences since it omits the HQC). It
prints rho and DW only for time series.
* Version C is what I'm inclined to favour at present.
* Version D is what Jack canvassed in a later patch.
I'll make a case for C, though I'm not totally committed to this.
First, I don't think it's necessary to print the sample size in
this table, since we always show that in the header for model
output. Second, it's obviously a matter of opinion whether it's
"more logical" to have closely related items (e.g. the F-stat and
its p-value) on the same row (as in Jack's B) or next to each
other vertically (as in C). I find that my eye tends to group
things vertically given this sort of presentation, so I have a
slight preference for C in this regard.
Thoughts?
Allin.
2:55 p.m.
My two cents goto C also.
Cheers
Talha
P.S.: I compiled the latest cvs but it still shows the old output.
Will the new one go into effect later?
On Fri, Nov 21, 2008 at 4:00 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
On Fri, 21 Nov 2008, Sven Schreiber wrote:
> Apart from the looks, I'm not sure why Akaike should be
> preferred over Hannan-Quinn in terms of econometrics. I'm sure
> AIC is good for some objectives, but AFAIK it is not consistent
> in the sense that it doesn't pick the true model even
> asymptotically, in contrast to Hannan-Quinn.
>
> Am I mixing things up here, or are these thoughts irrelevnant
> for the issue we're discussing right now? Otherwise I would tend
> to either keep all or ditch AIC.
I agree, the AIC is not all that great. But it's the best known
of the information criteria and if we dropped it I guess we would
have people saying, why don't you print the AIC?
Here's another go at this topic. I've put up
http://ricardo.ecn.wfu.edu/~cottrell/dougherty/table.html
showing a few variants for folks to consider.
* Version A is what I posted yesterday. It is closely (too
closely?) based on eviews and does not have a space for the HQC;
also it prints the Durbin-Watson statistic regardless of whether
or not the data are time series (but does not print rho). We
probably don't want this.
* Version B is what Jack suggested yesterday (although it doesn't
fully correspond with his preferences since it omits the HQC). It
prints rho and DW only for time series.
* Version C is what I'm inclined to favour at present.
* Version D is what Jack canvassed in a later patch.
I'll make a case for C, though I'm not totally committed to this.
First, I don't think it's necessary to print the sample size in
this table, since we always show that in the header for model
output. Second, it's obviously a matter of opinion whether it's
"more logical" to have closely related items (e.g. the F-stat and
its p-value) on the same row (as in Jack's B) or next to each
other vertically (as in C). I find that my eye tends to group
things vertically given this sort of presentation, so I have a
slight preference for C in this regard.
Thoughts?
Allin.
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
"Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment." - Benjamin Franklin (1706-1790)
--
3:34 p.m.
On Fri, 21 Nov 2008, Talha Yalta wrote:
My two cents goto C also.
Cheers
Talha
P.S.: I compiled the latest cvs but it still shows the old output.
Will the new one go into effect later?
To get the new-style output, run gretl with the environment
variable COMPACT_MIDDLE defined, as in
COMPACT_MIDDLE=1 gretl
(or "export COMPACT_MIDDLE=1" then run gretl)
Since there seems to be a consensus in favour of the compact
version, we'll make that the default once we've sorted out
exactly what it should look like.
Allin
3:45 p.m.
To get the new-style output, run gretl with the environment
variable COMPACT_MIDDLE defined, as in
COMPACT_MIDDLE=1 gretl
Thanks. When I run it, the indentations in my translated
output look
wrong. Is this because this is not finalized yet? Below is an example:
(you need to display it using a monospace font)
katsayı ölç. hata t-oranı p-değeri
---------------------------------------------------------
const −184,078 46,2620 −3,979 0,0016 ***
X 0,706408 0,00782749 90,25 1,42E-19 ***
Bağımlı değ ort. 3964,087 Bağımlı değ Ö.S. 489,6614
Örneklem büyüklüğü 15 Kalıntı kareler top 5349,390
R-kare 0,998406 Düzeltilmiş R-kare 0,998284
Bağlanım Ö.H. 20,28525 Log-olabilirlik -65,35923
Akaike ölçütü 134,7185 Schwarz ölçütü 136,1346
F(1, 13) 8144,534 P-değeri(F) 1,42e-19
rho -0,042748 Durbin-Watson 2,081830
Ö.S. ve Ö.H. ölçünlü sapma ve ölçünlü hatayı göstermektedir.
--
"Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment." - Benjamin Franklin (1706-1790)
--
5:22 p.m.
New subject: Some strings not marked for translation
Allin,
in the gretl graphic controller, there are some strings not marked for
translation:
- In "key position", the options "left top" ... "none"
- In the "lines" flap, the "type" options: "points" ...
"steps"
- In the "labels" flap, the "justification" options: "left",
"center", "right"
- In the "palette" flap, the sentence "These colors will be used unless
overridden by graph-specific choices"
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.et.bs.ehu.es
3:13 p.m.
On Fri, 21 Nov 2008, Allin Cottrell wrote:
On Fri, 21 Nov 2008, Sven Schreiber wrote:
> Apart from the looks, I'm not sure why Akaike should be
> preferred over Hannan-Quinn in terms of econometrics. I'm sure
> AIC is good for some objectives, but AFAIK it is not consistent
> in the sense that it doesn't pick the true model even
> asymptotically, in contrast to Hannan-Quinn.
Come on, AIC ain't THAT bad. Only, the probability of AIC picking the
right model doesn't go to 1 asymptotically. From this ponit of view, BIC
and HQC are superior. However, AIC has the merit of having been the first
widely used criterion, and I would think it has to stay in for
comparison/replication purposes. By the same logic, one would have to
ditch the Durbin-Watson statistic, Cochrane-Orcutt, the ADF test... and
the list could continue.
Here's another go at this topic. I've put up
http://ricardo.ecn.wfu.edu/~cottrell/dougherty/table.html
showing a few variants for folks to consider.
[...]
I'll make a case for C, though I'm not totally committed to
this.
First, I don't think it's necessary to print the sample size in
this table, since we always show that in the header for model
output.
Ok. However, I'd rather have the sample size together with the other
statistics, rather than in the header. But I don't have overly strong
feelings on this.
Second, it's obviously a matter of opinion whether it's
"more logical" to have closely related items (e.g. the F-stat and
its p-value) on the same row (as in Jack's B) or next to each
other vertically (as in C). I find that my eye tends to group
things vertically given this sort of presentation, so I have a
slight preference for C in this regard.
I agree that to some extent it's a matter of personal taste. As for me, I
tend to "go by row" instead. If I was just interested in (say) the
F-statistic, I'd find it more immediate to look for its row. That also
makes copy-n-paste easier, if needed.
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucchetti(a)univpm.it
http://www.econ.univpm.it/lucchetti
1:59 p.m.
On Fri, 21 Nov 2008, Riccardo (Jack) Lucchetti wrote:
On Fri, 21 Nov 2008, Allin Cottrell wrote:
> Second, it's obviously a matter of opinion whether it's
> "more logical" to have closely related items (e.g. the F-stat and
> its p-value) on the same row (as in Jack's B) or next to each
> other vertically (as in C). I find that my eye tends to group
> things vertically given this sort of presentation, so I have a
> slight preference for C in this regard.
I agree that to some extent it's a matter of personal taste. As
for me, I tend to "go by row" instead. If I was just interested
in (say) the F-statistic, I'd find it more immediate to look for
its row. That also makes copy-n-paste easier, if needed.
OK, here's one more go at it:
http://ricardo.ecn.wfu.edu/~cottrell/dougherty/table.html now
shows two variants of the table, one with primarily "vertical
association" of closely related stats and one with primarily
row-wise association.
At this point I don't have a strong preference. Any votes/further
suggestions?
Allin.
3:50 p.m.
I prefer the vertical association, but I think the "Mean of dependent var"
could be placed in the first column, like that:
Mean dependent var 2.962500 Sum squared resid 59.23226
S.E. of regression 0.473671 S.D. dependent var 0.615864
R-squared 0.419454 F(5, 264) 38.14888
Adjusted R-squared 0.408459 P-value(F) 2.14e-29
Log-likelihood -178.3244 Akaike criterion 368.6488
Schwarz criterion 390.2393 Hannan-Quinn 377.3186
rho 0.435670 Durbin-Watson 2.036421
I think the "mean" is the most fundamental statistics. What do you think?
Regards,
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
7:32 p.m.
On Sat, 22 Nov 2008, Allin Cottrell wrote:
OK, here's one more go at it:
http://ricardo.ecn.wfu.edu/~cottrell/dougherty/table.html now
shows two variants of the table, one with primarily "vertical
association" of closely related stats and one with primarily
row-wise association.
At this point I don't have a strong preference. Any votes/further
suggestions?
Actually, I now see a reason for preferring the "row-wise"
variant: this makes it easier to cut out the F-stat and its
p-value if they're not applicable for a given model.
Allin.
11:11 p.m.
On Sat, 22 Nov 2008, Allin Cottrell wrote:
On Sat, 22 Nov 2008, Allin Cottrell wrote:
> OK, here's one more go at it:
> http://ricardo.ecn.wfu.edu/~cottrell/dougherty/table.html now
> shows two variants of the table, one with primarily "vertical
> association" of closely related stats and one with primarily
> row-wise association.
>
> At this point I don't have a strong preference. Any votes/further
> suggestions?
Actually, I now see a reason for preferring the "row-wise"
variant: this makes it easier to cut out the F-stat and its
p-value if they're not applicable for a given model.
So, could the following (Allin's proposal with the adj. R^2 and the
F-statistic swapped) be a reasonable compromise?
Sum squared resid 59.23226 Mean dependent var 2.962500
S.E. of regression 0.473671 S.D. dependent var 0.615864
R-squared 0.419454 Adjusted R-squared 0.408459
F(5, 264) 38.14888 P-value(F) 2.14e-29
Log-likelihood -178.3244 Akaike criterion 368.6488
Schwarz criterion 390.2393 Hannan-Quinn 377.3186
rho 0.435670 Durbin-Watson 2.036421
I'd still like to have the information criteria on their own, but I
understand not many people share my preference.
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucchetti(a)univpm.it
http://www.econ.univpm.it/lucchetti
1:35 a.m.
On Sun, 23 Nov 2008, Riccardo (Jack) Lucchetti wrote:
On Sat, 22 Nov 2008, Allin Cottrell wrote:
> Actually, I now see a reason for preferring the "row-wise"
> variant: this makes it easier to cut out the F-stat and its
> p-value if they're not applicable for a given model.
So, could the following (Allin's proposal with the adj. R^2 and the
F-statistic swapped) be a reasonable compromise?
Sum squared resid 59.23226 Mean dependent var 2.962500
S.E. of regression 0.473671 S.D. dependent var 0.615864
R-squared 0.419454 Adjusted R-squared 0.408459
F(5, 264) 38.14888 P-value(F) 2.14e-29
Log-likelihood -178.3244 Akaike criterion 368.6488
Schwarz criterion 390.2393 Hannan-Quinn 377.3186
rho 0.435670 Durbin-Watson 2.036421
I like it!
I'd still like to have the information criteria on their own, but
I
understand not many people share my preference.
I can see a case for this, but if (at least part of) the object is
to economize on vertical lines, having the information criteria on
their own is less than ideal. In the format above, the criteria
take up 1.5 lines, but if they're to be presented on their own I
think 3 lines are needed: a space, a heading, then a line for the
three values.
Allin.
3:49 p.m.
On Sat, 22 Nov 2008, Allin Cottrell wrote:
> So, could the following (Allin's proposal with the adj. R^2
and the
> F-statistic swapped) be a reasonable compromise?
>
> Sum squared resid 59.23226 Mean dependent var 2.962500
> S.E. of regression 0.473671 S.D. dependent var 0.615864
> R-squared 0.419454 Adjusted R-squared 0.408459
> F(5, 264) 38.14888 P-value(F) 2.14e-29
> Log-likelihood -178.3244 Akaike criterion 368.6488
> Schwarz criterion 390.2393 Hannan-Quinn 377.3186
> rho 0.435670 Durbin-Watson 2.036421
I like it!
Anyone who doesn't? Speak now or forever hold your peace.
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucchetti(a)univpm.it
http://www.econ.univpm.it/lucchetti
4:15 p.m.
On Sun, 23 Nov 2008, Riccardo (Jack) Lucchetti wrote:
On Sat, 22 Nov 2008, Allin Cottrell wrote:
>> So, could the following (Allin's proposal with the adj. R^2 and the
>> F-statistic swapped) be a reasonable compromise?
>>
>> Sum squared resid 59.23226 Mean dependent var 2.962500
>> S.E. of regression 0.473671 S.D. dependent var 0.615864
>> R-squared 0.419454 Adjusted R-squared 0.408459
>> F(5, 264) 38.14888 P-value(F) 2.14e-29
>> Log-likelihood -178.3244 Akaike criterion 368.6488
>> Schwarz criterion 390.2393 Hannan-Quinn 377.3186
>> rho 0.435670 Durbin-Watson 2.036421
>
> I like it!
Anyone who doesn't? Speak now or forever hold your peace.
Well... third thoughts! There are quite a few contexts in which
we print somewhat modified regression statistics (for example, for
WLS or AR1 estimation). I presume it's desirable that we handle
as many such cases as we reasonably can in a manner consistent
with plain OLS results. And in that regard it's easiest if we go
all the way with your row-wise presentation. For instance, it's
handy to be able to skip a single row containing the mean and s.d.
of the dependent variable. So here's an example of what I'm
working with at present:
Mean dependent var 4.538837 S.D. dependent var 0.243346
Sum squared resid 1.617235 S.E. of regression 0.189575
R-squared 0.430985 Adjusted R-squared 0.405696
F(2, 45) 17.04203 P-value(F) 3.09e-06
Log-likelihood 13.26255 Akaike criterion -20.52509
Schwarz criterion -14.91149 Hannan-Quinn -18.40371
rho 0.018627 Durbin-Watson 1.960299
(with the last row omitted for non-time series).
Allin.
5:41 p.m.
I also like row wise a little better mainly because it is more logical
to me if we are to add or omit rows depending on the type of
estimation or the data set.
Cheers
Talha
On Sun, Nov 23, 2008 at 6:15 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
On Sun, 23 Nov 2008, Riccardo (Jack) Lucchetti wrote:
> On Sat, 22 Nov 2008, Allin Cottrell wrote:
>
> >> So, could the following (Allin's proposal with the adj. R^2 and the
> >> F-statistic swapped) be a reasonable compromise?
> >>
> >> Sum squared resid 59.23226 Mean dependent var 2.962500
> >> S.E. of regression 0.473671 S.D. dependent var 0.615864
> >> R-squared 0.419454 Adjusted R-squared 0.408459
> >> F(5, 264) 38.14888 P-value(F) 2.14e-29
> >> Log-likelihood -178.3244 Akaike criterion 368.6488
> >> Schwarz criterion 390.2393 Hannan-Quinn 377.3186
> >> rho 0.435670 Durbin-Watson 2.036421
> >
> > I like it!
>
> Anyone who doesn't? Speak now or forever hold your peace.
Well... third thoughts! There are quite a few contexts in which
we print somewhat modified regression statistics (for example, for
WLS or AR1 estimation). I presume it's desirable that we handle
as many such cases as we reasonably can in a manner consistent
with plain OLS results. And in that regard it's easiest if we go
all the way with your row-wise presentation. For instance, it's
handy to be able to skip a single row containing the mean and s.d.
of the dependent variable. So here's an example of what I'm
working with at present:
Mean dependent var 4.538837 S.D. dependent var 0.243346
Sum squared resid 1.617235 S.E. of regression 0.189575
R-squared 0.430985 Adjusted R-squared 0.405696
F(2, 45) 17.04203 P-value(F) 3.09e-06
Log-likelihood 13.26255 Akaike criterion -20.52509
Schwarz criterion -14.91149 Hannan-Quinn -18.40371
rho 0.018627 Durbin-Watson 1.960299
(with the last row omitted for non-time series).
Allin.
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
"Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment." - Benjamin Franklin (1706-1790)
--
6:02 p.m.
Am 23.11.2008 17:15, Allin Cottrell schrieb:
Mean dependent var 4.538837 S.D. dependent var 0.243346
Sum squared resid 1.617235 S.E. of regression 0.189575
R-squared 0.430985 Adjusted R-squared 0.405696
F(2, 45) 17.04203 P-value(F) 3.09e-06
Log-likelihood 13.26255 Akaike criterion -20.52509
Schwarz criterion -14.91149 Hannan-Quinn -18.40371
rho 0.018627 Durbin-Watson 1.960299
(with the last row omitted for non-time series).
Why actually have both the rho and the DW stat? (Assuming rho is the
first-order autocorrelation coeff, right?) I mean the DW stat only
provides additional information if the p-value is also given. And of
course there are no exact p-values for DW usually, and I would argue the
whole DW test is now obsolete and has been for some time, since we also
know that in many real-world circumstances like lagged endogenous
variables it isn't ok anyway. And a point-estimate-type information
about first-order autocorrelation is already contained in rho.
I know there are historical reasons for DW, but hey: change yes we can
... ;-) Sometimes it seems ridiculous to me that in every econometrics
course you have to talk a lot about the historical DW test just because
it always appears in the standard estimation output.
Next, an unrelated suggestion: What do you think about grouping the
adjusted R-squared and the info criteria together, since conceptually
they are more or less the same thing. Then of course you don't want to
separate the R-squared and the adjusted R-squared, so my suggestion
would be (apart from the DW issue before):
Mean dependent var 4.538837 S.D. dependent var 0.243346
Sum squared resid 1.617235 S.E. of regression 0.189575
F(2, 45) 17.04203 P-value(F) 3.09e-06
R-squared 0.430985 Log-likelihood 13.26255
Adjusted R-squared 0.405696 Akaike criterion -20.52509
Schwarz criterion -14.91149 Hannan-Quinn -18.40371
rho 0.018627 Durbin-Watson 1.960299
And a minor thing that can also be changed later: I think all
information criteria should have the criterion suffix and therefore it
should be "Hannan-Quinn crit.".
thanks,
sven
6:57 p.m.
On Sun, 23 Nov 2008, Sven Schreiber wrote:
Am 23.11.2008 17:15, Allin Cottrell schrieb:
>
> Mean dependent var 4.538837 S.D. dependent var 0.243346
> Sum squared resid 1.617235 S.E. of regression 0.189575
> R-squared 0.430985 Adjusted R-squared 0.405696
> F(2, 45) 17.04203 P-value(F) 3.09e-06
> Log-likelihood 13.26255 Akaike criterion -20.52509
> Schwarz criterion -14.91149 Hannan-Quinn -18.40371
> rho 0.018627 Durbin-Watson 1.960299
>
> (with the last row omitted for non-time series).
>
Why actually have both the rho and the DW stat? (Assuming rho is the
first-order autocorrelation coeff, right?)
Right.
I know there are historical reasons for DW, but hey: change yes we
can
... ;-) Sometimes it seems ridiculous to me that in every econometrics
course you have to talk a lot about the historical DW test just because
it always appears in the standard estimation output.
What I usually teach my students is that DW has just historical importance
and that if you want to test for autocorrelation the LM (Godfrey) test is
a much better tool. So, in principle I agree with you 100%. The only cases
(that I can think of right now) when I may want to actually see the DW
statistic are:
1) as always, for comparison/replication purposes
2) in the context of teaching what a spurious regression is (Granger &
Newbold's famous "rule of thumb")
3) in the context of cointegration tests, but IMHO nowadays Sargan and
Bhargava's idea is quite outdated too.
Hence, I wouldn't miss DW very much personally. If I could have it my way,
I'd show an LM test statistic (with df equal to the dataset's periodicity)
and the associated p-value instead. However, I can also see the case for
keeping DW in for those who prefer a more "traditional" approach.
Next, an unrelated suggestion: What do you think about grouping the
adjusted R-squared and the info criteria together, since conceptually
they are more or less the same thing.
No big objection here; R^2 and adj. R^2 fit together nicely IMHO for OLS,
but we have several estimation methods (eg probit) for which we don't have
the two, so it may make sense to split them. However, the log-likelihood
is the essential ingredient of the information criteria, so I'd like to
keep it where it is now.
And a minor thing that can also be changed later: I think all
information criteria should have the criterion suffix and therefore it
should be "Hannan-Quinn crit.".
Agreed. Once we're at it, I'd like to express my preference for the
"AIC",
"BIC" and "HQC" acronyms, which shouldn't necessarily be
translated.
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucchetti(a)univpm.it
http://www.econ.univpm.it/lucchetti
9:12 a.m.
El Sunday 23 November 2008 19:57:13 Riccardo (Jack) Lucchetti escribió:
On Sun, 23 Nov 2008, Sven Schreiber wrote:
> Am 23.11.2008 17:15, Allin Cottrell schrieb:
>> Mean dependent var 4.538837 S.D. dependent var 0.243346
>> Sum squared resid 1.617235 S.E. of regression 0.189575
>> R-squared 0.430985 Adjusted R-squared 0.405696
>> F(2, 45) 17.04203 P-value(F) 3.09e-06
>> Log-likelihood 13.26255 Akaike criterion -20.52509
>> Schwarz criterion -14.91149 Hannan-Quinn -18.40371
>> rho 0.018627 Durbin-Watson 1.960299
>>
>> (with the last row omitted for non-time series).
>
> Why actually have both the rho and the DW stat? (Assuming rho is the
> first-order autocorrelation coeff, right?)
I also don't see the necessity of rho (well, only for having an even number of
items here). As a former RATS user I miss, in the context of ARIMA models
(not for ols), the Ljung-Box statistic.
What I usually teach my students is that DW has just historical importance
and that if you want to test for autocorrelation the LM (Godfrey) test is
a much better tool. So, in principle I agree with you 100%. The only cases
(that I can think of right now) when I may want to actually see the DW
statistic are:
1) as always, for comparison/replication purposes
2) in the context of teaching what a spurious regression is (Granger &
Newbold's famous "rule of thumb")
3) in the context of cointegration tests, but IMHO nowadays Sargan and
Bhargava's idea is quite outdated too.
Hence, I wouldn't miss DW very much personally.
I vote for DW as well.
> Next, an unrelated suggestion: What do you think about grouping
the
> adjusted R-squared and the info criteria together, since conceptually
> they are more or less the same thing.
I prefer leaving the things as above. I consider the F() test may be
interpreted as a type of "significance test" for the R^2. Remember that
F=[R^2/(k-1)]/[(1-R^2)/(T-k)]
so I think it is natural that the F test goes inmediately after the R^2.
Agreed. Once we're at it, I'd like to express my preference
for the "AIC",
"BIC" and "HQC" acronyms, which shouldn't necessarily be
translated.
I also vote for AIC, BIC and HQC.
--
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.et.bs.ehu.es
9 p.m.
Hi!
I downloaded the latest snapshot of gretl for Windows (build date
2008-11-26) and I realized that the new model presentation layout was
abandoned in the Brazilian Portuguese translation, but not in English.
Best,
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
On Mon, Nov 24, 2008 at 7:12 AM, Ignacio Diaz-Emparanza
<ignacio.diaz-emparanza(a)ehu.es> wrote:
El Sunday 23 November 2008 19:57:13 Riccardo (Jack) Lucchetti
escribió:
> On Sun, 23 Nov 2008, Sven Schreiber wrote:
> > Am 23.11.2008 17:15, Allin Cottrell schrieb:
> >> Mean dependent var 4.538837 S.D. dependent var 0.243346
> >> Sum squared resid 1.617235 S.E. of regression 0.189575
> >> R-squared 0.430985 Adjusted R-squared 0.405696
> >> F(2, 45) 17.04203 P-value(F) 3.09e-06
> >> Log-likelihood 13.26255 Akaike criterion -20.52509
> >> Schwarz criterion -14.91149 Hannan-Quinn -18.40371
> >> rho 0.018627 Durbin-Watson 1.960299
> >>
> >> (with the last row omitted for non-time series).
> >
> > Why actually have both the rho and the DW stat? (Assuming rho is the
> > first-order autocorrelation coeff, right?)
>
I also don't see the necessity of rho (well, only for having an even number of
items here). As a former RATS user I miss, in the context of ARIMA models
(not for ols), the Ljung-Box statistic.
>
> What I usually teach my students is that DW has just historical importance
> and that if you want to test for autocorrelation the LM (Godfrey) test is
> a much better tool. So, in principle I agree with you 100%. The only cases
> (that I can think of right now) when I may want to actually see the DW
> statistic are:
>
> 1) as always, for comparison/replication purposes
> 2) in the context of teaching what a spurious regression is (Granger &
> Newbold's famous "rule of thumb")
> 3) in the context of cointegration tests, but IMHO nowadays Sargan and
> Bhargava's idea is quite outdated too.
>
> Hence, I wouldn't miss DW very much personally.
I vote for DW as well.
> > Next, an unrelated suggestion: What do you think about grouping the
> > adjusted R-squared and the info criteria together, since conceptually
> > they are more or less the same thing.
I prefer leaving the things as above. I consider the F() test may be
interpreted as a type of "significance test" for the R^2. Remember that
F=[R^2/(k-1)]/[(1-R^2)/(T-k)]
so I think it is natural that the F test goes inmediately after the R^2.
> Agreed. Once we're at it, I'd like to express my preference for the
"AIC",
> "BIC" and "HQC" acronyms, which shouldn't necessarily be
translated.
I also vote for AIC, BIC and HQC.
--
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.et.bs.ehu.es
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
9:25 p.m.
Hello,
The reason for this behaviour is very long string 'Logaritmo da
verossimilhanca'.
gretl prints the error message:
'Can't make compact model stats table -- the max length translated
string is 39 chars, should be < 23'.
Best regards,
Alexander Gedranovich
Henrique wrote:
Hi!
I downloaded the latest snapshot of gretl for Windows (build date
2008-11-26) and I realized that the new model presentation layout was
abandoned in the Brazilian Portuguese translation, but not in English.
Best,
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
9:40 p.m.
On Wed, 26 Nov 2008, Alexander B. Gedranovich wrote:
The reason for this behaviour is very long string 'Logaritmo da
verossimilhanca'.
gretl prints the error message:
'Can't make compact model stats table -- the max length translated
string is 39 chars, should be < 23'.
Yes, Alexander is exactly right. I've now put in place a fallback
mechanism: if any of the translated strings for the "compact"
model results table are longer than 22 characters, the table will
look quite bad, so we drop back to printing one statistic per
line. This takes up more space but should not look broken.
Please see the file README in the "po" subdirectory of the gretl
source.
Allin.
9:43 p.m.
Hi,
In that case we must abbreviate 'Logaritmo da verosimilhança' to 'Log.
da verosimilhança'.
:)
On Wed, Nov 26, 2008 at 11:25 PM, Alexander B. Gedranovich
<gedranovich(a)gmail.com> wrote:
Hello,
The reason for this behaviour is very long string 'Logaritmo da
verossimilhanca'.
gretl prints the error message:
'Can't make compact model stats table -- the max length translated
string is 39 chars, should be < 23'.
Best regards,
Alexander Gedranovich
Henrique wrote:
> Hi!
>
> I downloaded the latest snapshot of gretl for Windows (build date
> 2008-11-26) and I realized that the new model presentation layout was
> abandoned in the Brazilian Portuguese translation, but not in English.
>
> Best,
>
> Henrique C. de Andrade
> Doutorando em Economia Aplicada
> Universidade Federal do Rio Grande do Sul
> www.ufrgs.br/ppge
>
>
>
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
12:29 a.m.
Thank you guys! I going to correct the .po file right now!
Regards,
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
On Wed, Nov 26, 2008 at 7:43 PM, Hélio Guilherme <helioxentric(a)gmail.com> wrote:
Hi,
In that case we must abbreviate 'Logaritmo da verosimilhança' to 'Log.
da verosimilhança'.
:)
On Wed, Nov 26, 2008 at 11:25 PM, Alexander B. Gedranovich
<gedranovich(a)gmail.com> wrote:
> Hello,
>
> The reason for this behaviour is very long string 'Logaritmo da
> verossimilhanca'.
>
> gretl prints the error message:
> 'Can't make compact model stats table -- the max length translated
> string is 39 chars, should be < 23'.
>
> Best regards,
> Alexander Gedranovich
>
>
> Henrique wrote:
>> Hi!
>>
>> I downloaded the latest snapshot of gretl for Windows (build date
>> 2008-11-26) and I realized that the new model presentation layout was
>> abandoned in the Brazilian Portuguese translation, but not in English.
>>
>> Best,
>>
>> Henrique C. de Andrade
>> Doutorando em Economia Aplicada
>> Universidade Federal do Rio Grande do Sul
>> www.ufrgs.br/ppge
>>
>>
>>
>
> _______________________________________________
> Gretl-devel mailing list
> Gretl-devel(a)lists.wfu.edu
> http://lists.wfu.edu/mailman/listinfo/gretl-devel
>
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
6:58 p.m.
I already fix the problems:
BEFORE:
Média da var. dependente 0.004595
D.P. da var. dependente 0.041708
Soma dos resíduos quadrados 0.227919
E.P. da regressão 0.037625
R-quadrado 0.196207
R-quadrado ajustado 0.186222
F(2, 161) 19.65016
P-valor(F) 2.31e-08
Log da verossimilhança 306.7418
Critério de Akaike -607.4836
Critério de Schwarz -598.1840
Critério de Hannan-Quinn -603.7083
rô 0.365419
Estatística Durbin-Watson 1.260998
AFTER:
Média var. dependente 0.004595 D.P. var. dependente 0.041708
Soma resíd. quadrados 0.227919 E.P. da regressão 0.037625
R-quadrado 0.196207 R-quadrado ajustado 0.186222
F(2, 161) 19.65016 P-valor(F) 2.31e-08
Log da verossimilhança 306.7418 Critério de Akaike -607.4836
Critério de Schwarz -598.1840 Critério Hannan-Quinn -603.7083
rô 0.365419 Durbin-Watson 1.260998
What dou you think? I liked very much! But now, what I have to do? I
have the .po file and the .mo file (tested on my Mac and PC) working
fine.
Best,
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
On Wed, Nov 26, 2008 at 10:29 PM, Henrique <henrique.coelho(a)gmail.com> wrote:
Thank you guys! I going to correct the .po file right now!
Regards,
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
On Wed, Nov 26, 2008 at 7:43 PM, Hélio Guilherme <helioxentric(a)gmail.com> wrote:
> Hi,
>
> In that case we must abbreviate 'Logaritmo da verosimilhança' to 'Log.
> da verosimilhança'.
>
> :)
>
> On Wed, Nov 26, 2008 at 11:25 PM, Alexander B. Gedranovich
> <gedranovich(a)gmail.com> wrote:
>> Hello,
>>
>> The reason for this behaviour is very long string 'Logaritmo da
>> verossimilhanca'.
>>
>> gretl prints the error message:
>> 'Can't make compact model stats table -- the max length translated
>> string is 39 chars, should be < 23'.
>>
>> Best regards,
>> Alexander Gedranovich
>>
>>
>> Henrique wrote:
>>> Hi!
>>>
>>> I downloaded the latest snapshot of gretl for Windows (build date
>>> 2008-11-26) and I realized that the new model presentation layout was
>>> abandoned in the Brazilian Portuguese translation, but not in English.
>>>
>>> Best,
>>>
>>> Henrique C. de Andrade
>>> Doutorando em Economia Aplicada
>>> Universidade Federal do Rio Grande do Sul
>>> www.ufrgs.br/ppge
>>>
>>>
>>>
>>
>> _______________________________________________
>> Gretl-devel mailing list
>> Gretl-devel(a)lists.wfu.edu
>> http://lists.wfu.edu/mailman/listinfo/gretl-devel
>>
>
> _______________________________________________
> Gretl-devel mailing list
> Gretl-devel(a)lists.wfu.edu
> http://lists.wfu.edu/mailman/listinfo/gretl-devel
>
--
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
--
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
7:02 p.m.
On Thu, 27 Nov 2008, Henrique wrote:
I already fix the problems...
AFTER:
Média var. dependente 0.004595 D.P. var. dependente 0.041708
Soma resíd. quadrados 0.227919 E.P. da regressão 0.037625
R-quadrado 0.196207 R-quadrado ajustado 0.186222
F(2, 161) 19.65016 P-valor(F) 2.31e-08
Log da verossimilhança 306.7418 Critério de Akaike -607.4836
Critério de Schwarz -598.1840 Critério Hannan-Quinn -603.7083
rô 0.365419 Durbin-Watson 1.260998
What dou you think? I liked very much! But now, what I have to do? I
have the .po file and the .mo file (tested on my Mac and PC) working
fine.
Looks good to me! Just commit your current .po to CVS (or email
it to me and I'll commit it).
Allin.
7:16 p.m.
Allin,
I would like to commit the .po file for myself (to learn the
process), but I think I don't have access to change (upload) files on
gretl's CVS. So, I going to send you by e-mail, ok?
--
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
On Thu, Nov 27, 2008 at 5:02 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
On Thu, 27 Nov 2008, Henrique wrote:
> I already fix the problems...
> AFTER:
>
> Média var. dependente 0.004595 D.P. var. dependente 0.041708
> Soma resíd. quadrados 0.227919 E.P. da regressão 0.037625
> R-quadrado 0.196207 R-quadrado ajustado 0.186222
> F(2, 161) 19.65016 P-valor(F) 2.31e-08
> Log da verossimilhança 306.7418 Critério de Akaike -607.4836
> Critério de Schwarz -598.1840 Critério Hannan-Quinn -603.7083
> rô 0.365419 Durbin-Watson 1.260998
>
> What dou you think? I liked very much! But now, what I have to do? I
> have the .po file and the .mo file (tested on my Mac and PC) working
> fine.
Looks good to me! Just commit your current .po to CVS (or email
it to me and I'll commit it).
Allin.
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
5:04 p.m.
On Thu, 27 Nov 2008, Henrique wrote:
I would like to commit the .po file for myself (to learn the
process), but I think I don't have access to change (upload) files on
gretl's CVS. So, I going to send you by e-mail, ok?
Yes, fine. I'll set up CVS access for you if you can send me your
sourceforge username.
Allin.
6:56 p.m.
My sourceforge username is ch_andrade.
Regards,
Henrique
On Sat, Nov 29, 2008 at 3:04 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
On Thu, 27 Nov 2008, Henrique wrote:
> I would like to commit the .po file for myself (to learn the
> process), but I think I don't have access to change (upload) files on
> gretl's CVS. So, I going to send you by e-mail, ok?
Yes, fine. I'll set up CVS access for you if you can send me your
sourceforge username.
Allin.
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
9:24 p.m.
I just noticed that the abbreviations line doesn't show up in the
LaTeX version of the table.
Cheers
Talha
On Sat, Nov 29, 2008 at 8:56 PM, Henrique <henrique.coelho(a)gmail.com> wrote:
My sourceforge username is ch_andrade.
Regards,
Henrique
On Sat, Nov 29, 2008 at 3:04 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
> On Thu, 27 Nov 2008, Henrique wrote:
>
>> I would like to commit the .po file for myself (to learn the
>> process), but I think I don't have access to change (upload) files on
>> gretl's CVS. So, I going to send you by e-mail, ok?
>
> Yes, fine. I'll set up CVS access for you if you can send me your
> sourceforge username.
>
> Allin.
> _______________________________________________
> Gretl-devel mailing list
> Gretl-devel(a)lists.wfu.edu
> http://lists.wfu.edu/mailman/listinfo/gretl-devel
>
--
Henrique C. de Andrade
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
--
"Remember not only to say the right thing in the right place, but far
more difficult still, to leave unsaid the wrong thing at the tempting
moment." - Benjamin Franklin (1706-1790)
--
5867
days inactive
5876
days old
45 comments
8 participants
participants (8)
-
Alexander B. Gedranovich -
Allin Cottrell -
Henrique -
Hélio Guilherme -
Ignacio Diaz-Emparanza -
Riccardo (Jack) Lucchetti -
Sven Schreiber -
Talha Yalta