Minimum Gretl version
by Andreas Noack Jensen

I cannot set any of the last two counters in the package GUI to more than 9
and I would like to have the latest Gretl (1.9.12) as minimum requirement.
Best
Andreas Noack Jensen
8 years, 7 months

visible number of digits
by Sven Schreiber

Hi,
while trying to check whether some data were imported ok, I think I
found out that while in the "view values" (or similar, retranslated from
German now) window you can set the number display format, you cannot do
that in the "edit values" window, where it is perhaps even more
necessary. Or am I missing something?
And perhaps it would also be useful to have an option in the number
format dialog for "show all available digits / max precision" or
something like that.
(And BTW, I almost missed the button in the "view values" window because
the icon does not strike me as intuitive for indicating a number format
-- I know this is perhaps more a GTK thing, but still. Right now I'm
talking about the Windows version.)
Thanks,
sven
8 years, 8 months

parser problem
by Sven Schreiber

hi,
I was mixing up the parentheses and tried to access a single value of a
series like this:
x(1995:1) -- instead of x[1995:1]
The problem is that gretl did not complain, but instead seems to have
treated this as a lag expression and I got some weird results. Now,
AFAICS the expression '1995:1' is not a valid numeral that can be used
as a lag order, so I think gretl should give an error instead, no?
thanks,
sven
8 years, 8 months

Special volume about gretl in Journal of Statistical Software to celebrate gretl 2.0?
by andreas.rosenblad＠ltv.se

A few months ago, the Journal of Statistical Software invited ideas about
special volumes of the Journal of Statistical Software (JSS). See below.
Wouldn't it be a good idea to celebrate the release of gretl 2.0 with a
special volume in JSS about "Statistics and Econometrics in gretl"?
Best regards,
Andreas
jss-announce-bounces(a)lists.stat.ucla.edu skrev 2013-01-20 00:20:58 :
> Från: Jan de Leeuw <deleeuw(a)stat.ucla.edu>
> Till: JSS JSS <jss(a)stat.ucla.edu>
> Kopia: "jss-announce(a)stat.ucla.edu" <jss-announce(a)stat.ucla.edu>
> Datum: 2013-01-20 00:21
> Ärende: [JSS-Announce] time to start thinking about special volumes
> Sänt av: jss-announce-bounces(a)lists.stat.ucla.edu
>
> We welcome suggestions/proposals for special volumes. They are very
> popular and get a lot of citations. Guest editors invite
> contributions, handle preprocessing, handle reviews, and make
> decisions. JSS (i.e. Achim) still handles the post-processing. I
> could easily see special volumes on MCMC/Bayesian Computing, Cloud
> Computing / Big Memory, HPC, Literary Programming (Sweave and
> Friends), Clinical Trials, Spatial Statistics/Geostatistics, Time
> Series, Mixed Models, Foreign Function Interfaces, Bioinformatics,
> Independent Component Analysis, Julia and Statistics, L_1
> regression, .... Of course JSS is 80% R, but (as the state space
> special volume shows) using R is not in any sense required. We still
> have a special volume on optimization in the pipeline. Ten most
> recent volumes are
>
> Volume 49: Graphical User Interfaces for R
>
> Volume 48: Psychoco: Psychometric Computing in R
>
> Volume 45: Multiple Imputation
>
> Volume 44: Magnetic Resonance Imaging in R
>
> Volume 42: Political Methodology
>
> Volume 41: Statistical Software for State Space Methods
>
> Volume 38: Competing Risks and Multi-State Models
>
> Volume 27: Econometrics in R
>
> Volume 24: Statistical Modeling of Social Networks with "statnet"
>
> Volume 22: Ecology and Ecological Modelling in R
>
> _______________________________________________
> JSS-Announce mailing list
> JSS-Announce(a)lists.stat.ucla.edu
> http://lists.stat.ucla.edu/mailman/listinfo/jss-announce
8 years, 9 months

A short wish list
by andreas.rosenblad＠ltv.se

Dear gretl developers,
If possible, I would like to see the following features included in gretl
and available form the GUI, to make it more usable:
Univariate tests:
*Z-test of means for paired data
*McNemars test
*Kruskal-Wallis test
*Friedman's test
*Jonckheere-Terpstra test
For duration data:
*Kaplan-Meier curves plot
*log-rank test
*Cox regression
Regression models:
*conditional logistic (logit) regression
*generalized estimating equations
*generalized linear mixed models
Any chances of any of these being included?
Best regards,
Andreas Rosenblad
8 years, 9 months

Stata foreign woes
by Lee Adkins

I can't seem to make this work. Here is the example from the User's Guide
<hansl>
function matrix stata_reorder (matrix se)
scalar n = rows(se)
return se[n] | se[1:n-1]
end function
open data4-1
ols 1 0 2 3 --cluster=bedrms
matrix se = $stderr
foreign language=stata --send-data
regress price sqft bedrms, vce(cluster bedrms)
matrix vcv = e(V)
gretl_export vcv "vcv.mat"
end foreign
matrix stata_vcv = mread("@dotdir/vcv.mat")
stata_se = stata_reorder(sqrt(diag(stata_vcv)))
matrix check = se - stata_se
print check
<\hansl>
There is something fishy about gretl_export. I can't seem to get a matrix
written to the dotdir directory (or anywhere else). The data are created
and loaded into Stata, the regression runs in the background, but vcv.mat
is nowhere to be found. I'm using version Stata 12 on Windows.
Am I missing something obvious?
Lee
--
Lee Adkins
Professor of Economics
lee.adkins(a)okstate.edu
learneconometrics.com
8 years, 9 months

Re: [Gretl-devel] Stata foreign woes
by Gordon Hughes

Stata for Windows can handle forward slashes in file names. Indeed,
recommended practice is always to use them since Stata for other
systems usually can't handle back slashes.
BUT, you are not asking Stata to handle the file name. The adopath +
<pathname> command is simply adding an additional pathname to the
existing adopath which will, in turn, be handled by Windows. That is
why Windows is generating the error message.
So in this context you need to use the following
adopath + C:\myprogs
and not
adopath + C:/myprogs
Gordon Hughes
>Message: 2
>Date: Fri, 12 Apr 2013 09:33:36 -0500
>From: Lee Adkins <lee.adkins(a)okstate.edu>
>Subject: Re: [Gretl-devel] Stata foreign woes
>To: Gretl development <gretl-devel(a)lists.wfu.edu>
>Message-ID:
> <CA+uerCGj5XKb-k2uM1AwEapPdMGtAW=CAvuPgvepU_FQTkJJRQ(a)mail.gmail.com>
>Content-Type: text/plain; charset="iso-8859-1"
>
>Well, I'm not making much progress. The .ado file works manually, but I
>don't think Stata can find it when the foreign command is executed.
>
>Actually, gretl doesn't like the foreign language=stata in Windows at all.
>So, I don't think the problem is with gretl_export.
>
>In fact, this generates the error (Windows) "The System cannot find file
>specified."
>
><>
>foreign language=stata
>end foreign
></>
>
>It's possible that Stata is being confused by the forward slashes in the
>adopath +? This is the only statement that appears in the gretltmp.do.
> Stata normally looks for backslashes under windows. But I'm just
>guessing. Anyone else with Windows using Stata?
>
>Lee
>
>
>On Thu, Apr 11, 2013 at 3:14 PM, Lee Adkins <lee.adkins(a)okstate.edu> wrote:
>
> > Excellent. I'll give it a go. Stata is horrendous to program I think.
> > Or maybe I'm just not much of a programmer! Same result, though....
> >
> > Thanks,
> > Lee
> >
> >
> > On Thu, Apr 11, 2013 at 3:10 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
> >
> >> On Thu, 11 Apr 2013, Lee Adkins wrote:
> >>
> >> > I can't seem to make this work. Here is the example from
> >> > the User's Guide
> >> >
> >> > <hansl>
> >> > function matrix stata_reorder (matrix se)
> >> > scalar n = rows(se)
> >> > return se[n] | se[1:n-1]
> >> > end function
> >> >
> >> > open data4-1
> >> > ols 1 0 2 3 --cluster=bedrms
> >> > matrix se = $stderr
> >> >
> >> > foreign language=stata --send-data
> >> > regress price sqft bedrms, vce(cluster bedrms)
> >> > matrix vcv = e(V)
> >> > gretl_export vcv "vcv.mat"
> >> > end foreign
> >> >
> >> > matrix stata_vcv = mread("@dotdir/vcv.mat")
> >> > stata_se = stata_reorder(sqrt(diag(stata_vcv)))
> >> > matrix check = se - stata_se
> >> > print check
> >> > <\hansl>
> >> >
> >> > There is something fishy about gretl_export. I can't seem to get a
> >> matrix
> >> > written to the dotdir directory (or anywhere else). The data are
> >> created
> >> > and loaded into Stata, the regression runs in the background, but
> >> vcv.mat
> >> > is nowhere to be found. I'm using version Stata 12 on Windows.
> >>
> >> It works OK with Stata 12 on Linux. I don't have Stata for
> >> Windows so I'm not in a position to test. But the
> >> "gretl_export" command is implemented by gretl_export.ado
> >> (which is written into dotdir). It looks like this:
> >>
> >> <stata>
> >> program define gretl_export
> >> version 8.2
> >> local matrix `1'
> >> local fname `2'
> >> tempname myfile
> >> file open `myfile' using "`fname'", write text replace
> >> local nrows = rowsof(`matrix')
> >> local ncols = colsof(`matrix')
> >> file write `myfile' %8.0g (`nrows') %8.0g (`ncols') _n
> >> forvalues r=1/`nrows' {
> >> forvalues c=1/`ncols' {
> >> file write `myfile' %15.0e (`matrix'[`r',`c']) _n
> >> }
> >> }
> >> file close `myfile'
> >> end
> >> </stata>
> >>
> >> Perhaps you can experiment by running something like this
> >> manually.
> >>
> >> Note that there's no dotdir path in this function (sorry,
> >> "program"!), but the writing to dotdir is (or should be)
> >> achieved by gretl doing chdir() into dotdir before running the
> >> stata commands under "foreign" -- this being the only way we
> >> can control where stata writes its .log output.
> >>
> >> Allin
> >>
8 years, 9 months

gretl instances
by Allin Cottrell

A follow-up to Jack's posting at
http://lists.wfu.edu/pipermail/gretl-devel/2013-March/004371.html
The question Jack raised was about what happens when you
already have gretl running, and you double-click a file for
which gretl is registered as the handler (e.g. a .gdt data
file or .inp script file).
The OS defaults are: (a) for Linux and Windows you'd get the
file opened by a new gretl instance, but (b) on OS X (at least
with the new gtk-quartz build of gretl) you'd get the file
opened in the current gretl instance.
To provide choice on this on the OS X side I added (again this
is just in the gtk-quartz build) an option to open a new gretl
instance if you want.
Now on Linux and Windows I've added some functionality to
support choice. If you double-click on a gretl-type file this
will launch a new gretl instance, but that instance will now,
before showing itself, look around to see if gretl is already
running. If so, it'll put up a dialog box asking if you want a
new instance. And if you say "No" it'll pass a message to the
prior instance requesting that the file be opened, then exit.
Thi is implemented via the POSIX signals API on Linux, and the
Windows Messages API on win32.
This is in CVS and the Windows snapshot. Please test, and
report any breakage! Thanks.
Allin
8 years, 9 months

ADF test variations
by Allin Cottrell

As most readers will know, gretl offers a --test-down option for the
ADF test. To date (until recent CVS), the only form of "testing
down" that we offered was that proposed by Ng and Perron in the
mid-1990s. Namely, we start at the maximum lag order, and delete the
last lag (of the difference of the variable to be tested) so long as
it is not "significant" at the 10 percent level on a two-tailed
test.
The problem with this procedure is that (as later work by Ng and
Perron showed), it tends to over-parameterize the ADF regression,
hence reducing its power. A 2001 Econometrica article by these
authors suggested using a modified version of the AIC criterion to
optimize the lag order.
That's now implemented in gretl, via --test-down=MAIC (and for good
measure, we also offer the correspondingly modified Schwarz Bayesian
criterion, via --test-down=MBIC).
I think it would be a good idea to make the MAIC criterion the
default for "test-down". This would be backward incompatible, in the
sense that the same script, using the --test-down option with the
"adf" command, and with no additional parameter, would (in general)
produce different results under gretl >= 1.9.13 versus gretl <=
1.9.12. But you could still get the old behaviour under new gretl
using --test-down=tstat.
So my question to you: Does this sound OK? Any objections?
Allin
8 years, 9 months