Re: [Gretl-users] The "invcdf" function -Correction needed in Help?
by Alecos Papadopoulos
Thanks Sven , your explanations were helpful.
As regards the invcdf function, according to Help, the invcdf function
can be used for both the Binomial and the Poisson distributions. But it
indeed seems not to work for either, I tried it for various parameter
values.
So perhaps the Help should be corrected?
Alecos Papadopoulos
Athens University of Economics and Business, Greece
Department of Economics
cell:+30-6945-378680
fax: +30-210-8259763
skype:alecos.papadopoulos
On 16/10/2014 19:00, gretl-users-request(a)lists.wfu.edu wrote:
>> >Now, the function
>> >invcdf(B, prob, 1, 0.05)
>> >returns "NA"
> A CDF for discrete variables is not strictly increasing and thus not
> invertible. This could be the rationale for NA. Perhaps alternative
> conventions could be used for invcdf(), but you would have to argue that
> it's useful or necessary.
>
> hth,
> sven
12 years
Cointegration tests on many pairs of historic series
by Matteo Chiorboli
Dear Gretl users,
I need your help for running the Johansen test automatically on (n)
historic series, matched in pairs.
Substantially, I have 30 historic series, and I must test the cointegration
between all the possible pairs. Then, existing 435 possible pairs, I need
an automatic procedure to make this great number of Johansen tests.
Overall, I need a procedure that can run all possible cointegration tests
on a number of pairs of historic series given by the binomial coefficient
(combination), that in the specific case is n=30, k=2, then
[30!/(2!*(30-2)!]=435 pairs, and 435 Johansen tests.
Thank you for you help!
Matteo Chiorboli
12 years
Re: [Gretl-users] Bernoulli and the "critical" and "invcdf" functions
by Alecos Papadopoulos
Good afternoon.
I run gretl 1.9.92 64-bit for Windows.
For 0.05 < prob < 0.95
The function
critical(B, prob, 1, p=0.05)
returns the value "1".
As the Help says, the function "returns x such that P(X > x) = p". Since
number of trials is equal to 1, we just have a Bernoulli, with P(X=0) =
1-prob and P(X=1) =prob. So by returning "1" the function asserts that
P(X > 1) = 0.05 > 0
If the function returned "0" it would assert that P(X > 0) = 0.05 < prob
=P(X=1)
I take note of the asymmetry (in the first case, the function "sends
some probability to the right"), -can this be the rule that leads to the
function returning "1"?
Now, the function
invcdf(B, prob, 1, 0.05)
returns "NA"
As the Help says the function "returns x such that P(X ≤ x) = p"
So here a return of "1" would assert P(X =<1) = 0.05 < 1, while a return
of "0" would assert that P(X =<0) =0.05 < 1-Prob.
Here in both cases, returning a number, either 0 or 1, would
"under-signal" the true probability - so maybe this is the reason why
here we get "NA"?
It appears as a choice to prefer making the rejection of a null
hypothesis harder rather than easier... but most probably I am imagining
things.
Can somebody shed some light?
--
Alecos Papadopoulos
Athens University of Economics and Business, Greece
Department of Economics
cell:+30-6945-378680
fax: +30-210-8259763
skype:alecos.papadopoulos
12 years
crash in icon view
by Summers, Peter
I've come across another bug, this one causing a crash on Windows 7. If I open a data set (I've tried this with several), open the icon view, then double-click on either the "summary" or "correlations" icon, everything goes gray and I get a message saying "gretl.exe has stopped working. A problem caused the program to stop working correctly. Windows will close the program and notify you if a solution is available." This is v. 1.10.0cvs, built 10/9. The dataset info and notes icons open fine.
TIA,
PS
[cid:logoimg]
12 years
scatterplot bug
by Summers, Peter
Folks,
I think I've found a bug in the GUI scatterplot command. I generate a scatterplot, then edit it and change the default point marker (eg, from '+' to 'x'). I click apply to make the change, but when I try to change it back, nothing happens. This is on Windows 7, 64-bit cvs, v. 1.10.0, built 10/9.
PS
[cid:logoimg]
12 years
new in CVS
by Allin Cottrell
There are few small new things in gretl CVS and snapshots that I
haven't had time to document yet, but that people might like to try.
Here's a listing:
* All single-equation estimation commands now support a new option
--window (or just -w). The effect is that if such a command is
executed via the console or script in the GUI program, the estimates
are (besides being printed) displayed in a window of their own. This
is similar to the effect you could get previously by doing, for
example
foo <- ols y 0 X
foo.show
The differences are (a) that -w is much easier to type and (b) the
model is not automatically saved "as an icon", though you can do
that via the File menu in the model window if you want.
* The GUI command log (available via /Tools/Command log) now updates
itself in real time: if you keep it open you'll see the CLI
equivalent of GUI actions appear line by line. (Thanks to Jack for
suggesting this.)
* The GUI "console" now supports parenthesis matching.
* There's a new boolean "set" variable named "robust_z". It's "off"
by default, but if you do "set robust_z on" then whenever a model is
eatimated with the --robust option the "t-ratios" will be labeled
"z" and the associated p-values will be taken from the normal
distribution.
* RTF output (no doubt mostly of interest to users on Windows):
non-ASCII characters are now encoded using "\u<UC>" where <UC> is
the Unicode code-point (rather than attempting to recode to the
legacy Microsoft per-country "code-pages"). This should mean that
international text and symbols come out right in a greater variety
of conditions.
Allin Cottrell
12 years
Re: [Gretl-users] new in cvs
by cociuba mihai
>
> > One possible issue associated with this suggestion is that this
> > increases further the number of open windows in gretl. The need to
> > constantly move across quite independent windows is something that
> > sometimes I find really annoying. I appreciate the reasons behind this
> > strategy and I am not suggesting to switch to a completely different
> > setting. A possible escape and, according to me, a way to have the best
> > of both worlds is giving users the possibility to join/divide the
> > frames. I am thinking about something similar to what you can do in
> > Blender:
> >
> http://wiki.blender.org/index.php/Doc:2.6/Manual/Interface/Window_system/...
> > If you introduce also the possibility to save a given arrangement, this
> > will give the possibility to easily customize the workspace in a very
> > flexible way, without harming users with small screens. This way a user
> > can also easily recreate any frame arrangement he/she is more familiar
> > with (rstudio, stata, matlab,eviews...). (Any small change that reduces
> > switching costs should be welcome if we want to increase the number of
> > users.)
>
> This has been discussed many times. I personally don't want the gretl GUI
> to become a window manager, and I think Allin feels the same way too. By
> the way, have you tried Alt-PgUp/Alt-PgDn?
>
>
>
> Dear developers,
I think that if Gretl will be used on tablet (maybe even phones) then the
problem on multiple window should be analyzed in depth.
Why:
- the workflow is different on a tablet versus your PC , and even if there
are some great keyboards it's kind a difficult to use shortcuts
- allowing user to have the possibility of pinning the windows would be
great (attached you can find a screenshot with Rstudio and Gretl running in
a web browser)
Possible solution:
- an approach similar to Gimp (single window as an option)
http://askubuntu.com/questions/136992/how-to-make-gimp-run-as-a-single-wi...
- or reducing the number of new windows - output and graph one window
different tabs, logs and script one window (different tabs) etc.
Mihai
12 years
