9:23 a.m.
I have a bug report and a feature request:
1)- When I open a saved model (OLS) based on a sub-sample, I see that,
when running various tests, the original subsample is not always
remebered. For example, I have a model saved as an icon which uses the
variables 1-90. When I open the model with the full range restored,
White's test, BP and Chow tests for example remember the subsample
while RESET and normality of residual tests don't. Here, RESET test
returns 1.8e+308 while the normality of the residual window reports
using the observations 1-350 instead of 1-90 (although the results
seem to be correct).
Moreover, instead of full range, if I use another range, say 91-150,
White's tests and others also seem to return misleading output or
error messages such as "missing values encountered" or "Insufficient
data to build frequency distribution for variable uhat7." I think it
is very important that all tests avaliable for different models are
checked for this type of problem. One more thing to consider here is
that it is possible (as probably should be) to change the sample range
while a model window is open and doing this also causes the above
errors so this situation should also be taken into account.
2)- It would be really neat if there is an option for the model table
that makes it possible to set the number of digits or decimal places
for the coefficients as well as se's and whether t stats will be shown
etc. For example, I have a table with 8 models which could be much
smaller if I show 3 digits.
Also related, the LaTeX output for the model table uses hardcoded
\footnotesize commands for each standard error value. If I need to
make the whole table smaller, these remain unchanged resulting in a
weird looking table. Of course I can fix this by using the find and
replace command of the editor but I think it would be neat if there
was an option for the model table for setting table size, keeping
everything nice and proportional.
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)
--
3:23 p.m.
On Fri, 11 Sep 2009, Talha Yalta wrote:
I have a bug report and a feature request:
1)- When I open a saved model (OLS) based on a sub-sample, I see that,
when running various tests, the original subsample is not always
remebered. For example, I have a model saved as an icon which uses the
variables 1-90. When I open the model with the full range restored,
White's test, BP and Chow tests for example remember the subsample
while RESET and normality of residual tests don't. Here, RESET test
returns 1.8e+308 while the normality of the residual window reports
using the observations 1-350 instead of 1-90 (although the results
seem to be correct).
Moreover, instead of full range, if I use another range, say
91-150, White's tests and others also seem to return misleading
output or error messages... One more thing to consider here is
that it is possible (as probably should be) to change the sample
range while a model window is open and doing this also causes
the above errors so this situation should also be taken into
account.
Could you please give a step-by-step account of how to provoke
this problem? There is supposed to be a guard in place against
this sort of thing and in my experience the guard seems
effective. I guess it's breaking down in some circumstances but
I'm not sure how.
Here's an example:
* Open Ramanathan data4-10
* Use "/Sample/Restrict based on criterion" to restrict the
sample to WHITE > 0.5
* Estimate a model via OLS and save it as an icon
* Save the session to file
* Restart gretl and open the session file
* Choose "/Sample/Restore full range"
* Double-click on the saved model in the icon window to
display the model
* Click on the Tests menu in the model window
I get the message: "model is subsampled, dataset is not", and the
Tests menu is deactivated (grayed out), along with various other
items in the model-window menus. This is as it should be (though
it might be nice if there were an option here to re-establish the
sub-sample on which the particular model was estimated).
* Now do "/Sample/Restrict" again and give the criterion
MEMNEA < 0.9, producing a different subsample
* View the original model again and try the Tests menu
I now get "model and dataset subsamples not the same", and again
the Tests menu is deactivated.
So what I'm interested in is what you're doing that is getting
around this automatic guard.
Allin
8:02 p.m.
On Fri, 11 Sep 2009, Allin Cottrell wrote:
On Fri, 11 Sep 2009, Talha Yalta wrote:
> 1)- When I open a saved model (OLS) based on a sub-sample, I see that,
> when running various tests, the original subsample is not always
> remebered...
Could you please give a step-by-step account of how to provoke
this problem? There is supposed to be a guard in place against
this sort of thing and in my experience the guard seems
effective. I guess it's breaking down in some circumstances but
I'm not sure how.
Never mind, I figured out what the problem case is. A fix is in
progress.
Allin.
12:40 a.m.
On Fri, 11 Sep 2009, Allin Cottrell wrote:
On Fri, 11 Sep 2009, Talha Yalta wrote:
> 1)- When I open a saved model (OLS) based on a sub-sample, I see that,
> when running various tests, the original subsample is not always
> remebered...
I think this should all be OK now (in CVS and the various
snapshots).
For the record, here's the issue: you estimate model A, then
change the data sample, then run a diagnostic test of some kind on
model A. This is not a situation that would arise in a sane gretl
script, but it might well arise in the context of a saved session,
so we have to deal with it correctly.
For a long time now, we have dealt with it correctly if the sample
on which model A was estimated and the current sample differ in
certain specific ways, viz.
* The current sample is a non-contiguous subset of the full set of
observations, but the model was estimated on the full dataset, or
vice versa.
* The current sample and the sample on which the model was
estimated are different non-contiguous subsets of the full
dataset.
In these cases diagnostic tests are disabled.
But that leaves the case where the sample on which the model was
estimated and the current sample range differ only in respect of
where the "start" and "end" markers for the sample are situated.
In this case we ought to be able to reset the current sample range
temporarily to match that associated with the model, run the test,
then put the start and end back where they were.
We have done this correctly for many of the diagnostic tests that
gretl offers, but, as Talha noticed, not for all. I think we've
now plugged that gap. Please let me know if you see any cases that
seem to violate this requirement.
Allin
11:30 a.m.
Thanks very much for rapidly fixing the problem regarding some of the
diagnostic tests from a saved model based on a different sub-sample.
The new reformat options for the model table also looks very nice. Is
it just me or is gretl slowly but surely becoming the best tool for
doing econometrics?
Cheers
A. Talha Yalta
On Sat, Sep 12, 2009 at 3:40 AM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
On Fri, 11 Sep 2009, Allin Cottrell wrote:
> On Fri, 11 Sep 2009, Talha Yalta wrote:
>
> > 1)- When I open a saved model (OLS) based on a sub-sample, I see that,
> > when running various tests, the original subsample is not always
> > remebered...
I think this should all be OK now (in CVS and the various
snapshots).
For the record, here's the issue: you estimate model A, then
change the data sample, then run a diagnostic test of some kind on
model A. This is not a situation that would arise in a sane gretl
script, but it might well arise in the context of a saved session,
so we have to deal with it correctly.
For a long time now, we have dealt with it correctly if the sample
on which model A was estimated and the current sample differ in
certain specific ways, viz.
* The current sample is a non-contiguous subset of the full set of
observations, but the model was estimated on the full dataset, or
vice versa.
* The current sample and the sample on which the model was
estimated are different non-contiguous subsets of the full
dataset.
In these cases diagnostic tests are disabled.
But that leaves the case where the sample on which the model was
estimated and the current sample range differ only in respect of
where the "start" and "end" markers for the sample are situated.
In this case we ought to be able to reset the current sample range
temporarily to match that associated with the model, run the test,
then put the start and end back where they were.
We have done this correctly for many of the diagnostic tests that
gretl offers, but, as Talha noticed, not for all. I think we've
now plugged that gap. Please let me know if you see any cases that
seem to violate this requirement.
Allin
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
--
“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)
--
9:58 a.m.
I would like to express my concern regarding the recent modifications
to gretl fixing some of the diagnostic tests from a saved model based
on a different sub-sample. The current version of gretl indeed does
not produce erronious output in these tests because, if one opens a
previously estimated model, what gretl does is to give such error
messages as "dataset is subsampled, model is not" or "model and
dataset subsamples not the same" and to disable the Tests menu
altogether. IMHO this way of disabling functionality is not a good
solution.
What would be really neat and to be expected of a high quality program
that is gretl is that the program remembers the subsample for each
model saved as an icon and automatically adjusts the subsample when
these models are revisited.
I understand that this may not be too easy to implement but this is
really the only correct solution. I think the whole concept of
sessions is based on the expectation of such behavior. If I have a
bunch of saved models, I expect to be able to revisit any of them
later, run some tests and compare results without difficulty. The
current fix makes this unnecessarily difficult. Moreover, the current
behavior is also problematic because the subsamples can be random. If
I have a saved model using a random subsample and want to revisit or
compare results from a test, the program basically tells me that this
is not possible anymore, rendering the saved model useless.
Maybe this issue needs some rethinking?
Cheers
A. Talha Yalta
--
“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)
--
3:11 p.m.
On Thu, 17 Sep 2009, Talha Yalta wrote:
I would like to express my concern regarding the recent
modifications
to gretl fixing some of the diagnostic tests from a saved model based
on a different sub-sample. The current version of gretl indeed does
not produce erronious output in these tests because, if one opens a
previously estimated model, what gretl does is to give such error
messages as "dataset is subsampled, model is not" or "model and
dataset subsamples not the same" and to disable the Tests menu
altogether. IMHO this way of disabling functionality is not a good
solution.
I agree it's not ideal, but I'd like to clarify one thing: this is
the way it has always been. The recent fix didn't touch this
issue; it concerned the case where the current dataset sample and
the model sample differ _only_ in respect of the starting and/or
ending point of the sample range. Most model tests already
incorporated a correct workaround for that situation, and the
recent fix just involved ensuring that all tests use the same
workaround.
Allin.
1:30 p.m.
I agree it's not ideal, ...
I think it is far from being ideal. My understanding of the concept of
a session based work model is one where the users are able to estimate
different models and later revisit them and to all sorts of
comparisons and tests. There is little advantage of saving a model if
the user will not be able to perform any tests, draw and graphs or
save the fitted values etc.
Currently, what gretl wants is the user to manually return back to the
original subsample before performing such operations but this may not
be even possibe as in the case of models based on a random subsample.
Plus, I notice that the "Modify Model..." function under the Edit menu
is not affected by this behavior. So if one open a model, say using
the full range, and if the data is currently subsampled, after
modifying the model by adding or subtracting various variables, what
he gets will be a model based on a different subsample.
The current situation can also be annoying because, OK, say I opened a
model using a different subsample and tried to run a test. Gretl
warns me that the model is subsampled differently and makes the Tests
and Graphs menu ghosted. (Why are not the menus ghosted as soon as I
open the model window is a different question here.) So I go ahead and
manually adjust the subsample but what happens now is that those menus
are still ghosted so I need to first close the window and reopen it
just because they will not be ghosted anymore.
In conclusion, I have a suggestion: Could you please make it such that
whenever a model is saved as an icon, gretl internally saves the
observation numbers of the subsample as a vector? This will make it
possible that when the user opens a saved model or activates an open
model window, gretl first checks whether the current subsample matches
with those of the model and opens up a dialog box asking if the user
wants to return to the subsample used by the model. If the user say
"yes" the subsample is automatically adjusted. If the user says "no,"
then the Tests and Graphs menus are diasabled along with the "Modify
model" option.
Could you please consider this?
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)
--
2:36 p.m.
On Fri, 18 Sep 2009, Talha Yalta wrote:
> I agree it's not ideal, ...
I think it is far from being ideal. My understanding of the concept of
a session based work model is one where the users are able to estimate
different models and later revisit them and to all sorts of
comparisons and tests. There is little advantage of saving a model if
the user will not be able to perform any tests, draw and graphs or
save the fitted values etc.
Currently, what gretl wants is the user to manually return back to the
original subsample before performing such operations but this may not
be even possibe as in the case of models based on a random subsample.
Plus, I notice that the "Modify Model..." function under the Edit menu
is not affected by this behavior. So if one open a model, say using
the full range, and if the data is currently subsampled, after
modifying the model by adding or subtracting various variables, what
he gets will be a model based on a different subsample.
I'll come back to that point below.
The current situation can also be annoying because, OK, say I opened
a
model using a different subsample and tried to run a test. Gretl
warns me that the model is subsampled differently and makes the Tests
and Graphs menu ghosted. (Why are not the menus ghosted as soon as I
open the model window is a different question here.) ...
And here's the answer: because we don't know, at the time when the
model window is opened, what the sample will be when you actually
click on the relevant menu items. Part of the "non-linearity" of
a point-and-click session (convenient, up to a point, for the user
but quite tiresome for the programmer) is that the sample can be
changed while a model is displayed.
So I go ahead and manually adjust the subsample but what happens
now is that those menus are still ghosted so I need to first
close the window and reopen it just because they will not be
ghosted anymore.
If you get the matching subsample right the menus will become
active again. E.g.
open data4-1
restrict the sample using "bedrms=4"
run an OLS
display the model
restore the full sample range
try the menus in the model window: some items are inactive
again restrict the sample using "bedrms=4"
revisit the model menus: now they're all active
In conclusion, I have a suggestion: Could you please make it
such that whenever a model is saved as an icon, gretl internally
saves the observation numbers of the subsample as a vector?
We've always done that: that's how we can tell whether or not the
current sample differs from the one on which the model is based.
This will make it possible that when the user opens a saved
model or activates an open model window, gretl first checks
whether the current subsample matches with those of the model
and opens up a dialog box asking if the user wants to return to
the subsample used by the model. If the user say "yes" the
subsample is automatically adjusted. If the user says "no," then
the Tests and Graphs menus are diasabled along with the "Modify
model" option.
That's one way of handling the issue. What is now (newly)
implemented in CVS/snapshot is that most tests/graphs are enabled
regardless: if need be, we recreate the model's dataset on the fly
and do the tests/graphs using that. (This facility may still have
some rough edges.)
However, there are a couple of things that can't be handled in
that way, notably forecasts and saving series (e.g. residuals) to
the dataset. So having an option to restore the model's sample as
the current sample might be a good addition.
As for the "Modify model" item, I'm not sure there's any definite
"right behavior", in relation to the sample. But having the
option you mention might take care of that.
Allin.
8:51 p.m.
On Fri, 18 Sep 2009, Allin Cottrell wrote:
On Fri, 18 Sep 2009, Talha Yalta wrote:
> ...[W]hen the user opens a saved model or activates an open
> model window, gretl first checks whether the current subsample
> matches with those of the model and opens up a dialog box
> asking if the user wants to return to the subsample used by
> the model. If the user say "yes" the subsample is
> automatically adjusted. If the user says "no," then the Tests
> and Graphs menus are diasabled along with the "Modify model"
> option.
That's now the case, except that even if you say "No" to resetting
the sample, most items in the model menus will work OK.
Allin Cottrell
7:26 p.m.
On Fri, Sep 18, 2009 at 11:51 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
On Fri, 18 Sep 2009, Allin Cottrell wrote:
> On Fri, 18 Sep 2009, Talha Yalta wrote:
> > ...[W]hen the user opens a saved model or activates an open
> > model window, gretl first checks whether the current subsample
> > matches with those of the model and opens up a dialog box
> > asking if the user wants to return to the subsample used by
> > the model. If the user say "yes" the subsample is
> > automatically adjusted. If the user says "no," then the Tests
> > and Graphs menus are diasabled along with the "Modify model"
> > option.
That's now the case, except that even if you say "No" to resetting
the sample, most items in the model menus will work OK.
The newly implemeted behavior sounds outstanding, simply outstanding.
I will be away from computers for a few days but I cannot wait to
install and test the latest updates once I return Wednesday morning.
Thanks very very much for giving one more thing to brag about while
demonstrating gretl to the uninitiated :-)
Cheers
A. Talha Yalta
--
“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)
--
7:58 a.m.
I have been testing the improvements for model windows under changing
subsampling options. Thanks to Allins recent work, this functionality
looks very nice and works smoothly in general. I also have some
suggestions and bugs to report:
1)- In the Edit menu for the model window, it could be useful to add
an option such as "restore model sample" to quickly reverting to the
subsample used by that model.
2)- I understand disabling forecasts but don't see why gretl disables
saving the residuals, fitted values etc. from a model just because the
current subsample is different. I think it is perfectly OK to save the
residuals based on the model subsample with the outside observations
having missing values After all, this is what gretl does when the
current subsample happens to match the model subsample.
3)- A Bug: I have 2 models one is with n=100 and based on a dummy,
while the other is using a random subsample with n=200. After
estimating the second model, I return to the first one and gretl asks
me if I want to restore the sample. I say yes and now the sample size
is 60 instead of 100 (probably because gretl took the observations
only from the active random subsample of n=200 instead of the full
sample). The sample size keeps getting smaller as I open more model
windows and say yes to restoring the original sample! What is more, if
I now choose restore the full range from gretl's main window, the full
range is 60 instead of the full sample with n=400. The Set range
option also does not allow me to choose 1 to 400.
4)- The subsampling info at the bottom of the gretl main window could
use some improvements. For example, when the sub sample is based on a
dummy, instead of just saying "current sample n = 100" it would be
nice if gretl said something like: "current sample based on dummy
<varname> (n = 100)" (The same thing for other subsampling options
such as random subsamples etc.)
5)- I think the above improvement would be extremely useful for the
model output as well. For example, currently when I estimate an OLS
model based on a random subsample with n=200, the model output says:
"Model 1: OLS, using observations 1-200" which does not give any info
about the subsampling employed and giving the misleading impression
that the first 200 observations were used.
I also have some other small bugs to report:
6)- Opening a new data set, all open windows are closed except for the
gnuplot graphs.
7)- In the model specification window, when the "Set as default"
option is checked, it becomes unchecked the next time (although the
dependent variable is set correctly and as expected)
8)- My system is Turkish and when I use gretl in English, the legend
for the OLS fitted line still uses commas as the decimal seperator.
9)- Also, when using gretl in Turkish, entering formula for a new line
I need to use dot as the decimal seperator. If I use the comma sign in
the formula, gretl draws a wrong line and a differently wrong legend.
I think this issue was fixed a while ago.
These are all I can report right now. Again thanks for the great job
on this front.
Cheers
A. Talha Yalta
On Sat, Sep 19, 2009 at 10:26 PM, Talha Yalta <talhayalta(a)gmail.com> wrote:
On Fri, Sep 18, 2009 at 11:51 PM, Allin Cottrell
<cottrell(a)wfu.edu> wrote:
>
> On Fri, 18 Sep 2009, Allin Cottrell wrote:
>
>> On Fri, 18 Sep 2009, Talha Yalta wrote:
>> > ...[W]hen the user opens a saved model or activates an open
>> > model window, gretl first checks whether the current subsample
>> > matches with those of the model and opens up a dialog box
>> > asking if the user wants to return to the subsample used by
>> > the model. If the user say "yes" the subsample is
>> > automatically adjusted. If the user says "no," then the Tests
>> > and Graphs menus are diasabled along with the "Modify model"
>> > option.
>
> That's now the case, except that even if you say "No" to resetting
> the sample, most items in the model menus will work OK.
The newly implemeted behavior sounds outstanding, simply outstanding.
I will be away from computers for a few days but I cannot wait to
install and test the latest updates once I return Wednesday morning.
Thanks very very much for giving one more thing to brag about while
demonstrating gretl to the uninitiated :-)
Cheers
A. Talha Yalta
--
“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)
--
--
“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)
--
5:19 p.m.
On Thu, 24 Sep 2009, Talha Yalta wrote:
I have been testing the improvements for model windows under
changing
subsampling options. Thanks to Allins recent work, this functionality
looks very nice and works smoothly in general. I also have some
suggestions and bugs to report:
1)- In the Edit menu for the model window, it could be useful to add
an option such as "restore model sample" to quickly reverting to the
subsample used by that model.
Could be nice, not very high priority.
2)- I understand disabling forecasts but don't see why gretl
disables
saving the residuals, fitted values etc. from a model just because the
current subsample is different. I think it is perfectly OK to save the
residuals based on the model subsample with the outside observations
having missing values After all, this is what gretl does when the
current subsample happens to match the model subsample.
This is problematic if the model was estimated on a subsample and
the dataset is also currently subsampled, but in a different way
from the model. But some possibilities may be worth exploring.
3)- A Bug: I have 2 models one is with n=100 and based on a dummy,
while the other is using a random subsample with n=200. After
estimating the second model, I return to the first one and gretl asks
me if I want to restore the sample. I say yes and now the sample size
is 60 instead of 100 (probably because gretl took the observations
only from the active random subsample of n=200 instead of the full
sample)...
Thanks for the report; fixed in CVS.
4)- The subsampling info at the bottom of the gretl main window
could
use some improvements. For example, when the sub sample is based on a
dummy, instead of just saying "current sample n = 100" it would be
nice if gretl said something like: "current sample based on dummy
<varname> (n = 100)" (The same thing for other subsampling options
such as random subsamples etc.)
Too much information to print at the bottom of the window. I've
added a menu item /Sample/Show status which shows some more detail
than the bottom-of-the-screen line. It doesn't yet show a
sampling criterion or anything of that sort -- that could maybe be
added later (it would require recording more dataset information
than we currently record).
I also have some other small bugs to report:
6)- Opening a new data set, all open windows are closed except for the
gnuplot graphs.
OK, that could cause trouble in some cases. The graph windows
should now close too.
7)- In the model specification window, when the "Set as
default"
option is checked, it becomes unchecked the next time (although the
dependent variable is set correctly and as expected)
OK, cosmetic. Will be fixed at some point.
8)- My system is Turkish and when I use gretl in English, the legend
for the OLS fitted line still uses commas as the decimal seperator.
I'm not in a position to test that. Do other people see that too?
(I.e. you're in a locale that uses the decimal comma, but you use
the GUI selector to run in English: what decimal separator
character do you see?)
9)- Also, when using gretl in Turkish, entering formula for a new
line
I need to use dot as the decimal seperator. If I use the comma sign in
the formula, gretl draws a wrong line and a differently wrong legend.
I think this issue was fixed a while ago.
I'm pretty sure that's as it always has been. The string is fed
straight to gnuplot and you have to use '.' as decimal separator.
Allin.
11:15 p.m.
Allin Cottrell wrote:
On Thu, 24 Sep 2009, Talha Yalta wrote:
> I have been testing the improvements for model windows under changing
> subsampling options. Thanks to Allins recent work, this functionality
> looks very nice and works smoothly in general. I also have some
> suggestions and bugs to report:
>
> 1)- In the Edit menu for the model window, it could be useful to add
> an option such as "restore model sample" to quickly reverting to the
> subsample used by that model.
>
Could be nice, not very high priority.
> 2)- I understand disabling forecasts but don't see why gretl disables
> saving the residuals, fitted values etc. from a model just because the
> current subsample is different. I think it is perfectly OK to save the
> residuals based on the model subsample with the outside observations
> having missing values After all, this is what gretl does when the
> current subsample happens to match the model subsample.
>
This is problematic if the model was estimated on a subsample and
the dataset is also currently subsampled, but in a different way
from the model. But some possibilities may be worth exploring.
> 3)- A Bug: I have 2 models one is with n=100 and based on a dummy,
> while the other is using a random subsample with n=200. After
> estimating the second model, I return to the first one and gretl asks
> me if I want to restore the sample. I say yes and now the sample size
> is 60 instead of 100 (probably because gretl took the observations
> only from the active random subsample of n=200 instead of the full
> sample)...
>
Thanks for the report; fixed in CVS.
> 4)- The subsampling info at the bottom of the gretl main window could
> use some improvements. For example, when the sub sample is based on a
> dummy, instead of just saying "current sample n = 100" it would be
> nice if gretl said something like: "current sample based on dummy
> <varname> (n = 100)" (The same thing for other subsampling options
> such as random subsamples etc.)
>
Too much information to print at the bottom of the window. I've
added a menu item /Sample/Show status which shows some more detail
than the bottom-of-the-screen line. It doesn't yet show a
sampling criterion or anything of that sort -- that could maybe be
added later (it would require recording more dataset information
than we currently record).
Please do not forget to mark it for translation ;).
> I also have some other small bugs to report:
>
> 6)- Opening a new data set, all open windows are closed except for the
> gnuplot graphs.
>
OK, that could cause trouble in some cases. The graph windows
should now close too.
> 7)- In the model specification window, when the "Set as default"
> option is checked, it becomes unchecked the next time (although the
> dependent variable is set correctly and as expected)
>
OK, cosmetic. Will be fixed at some point.
> 8)- My system is Turkish and when I use gretl in English, the legend
> for the OLS fitted line still uses commas as the decimal seperator.
>
I'm not in a position to test that. Do other people see that too?
(I.e. you're in a locale that uses the decimal comma, but you use
the GUI selector to run in English: what decimal separator
character do you see?)
I tested this and when in English the coeficients appear with decimal point.
pt_PT cvs version under Ubunto (Acer Aspire One :))
> 9)- Also, when using gretl in Turkish, entering formula for a new
line
> I need to use dot as the decimal seperator. If I use the comma sign in
> the formula, gretl draws a wrong line and a differently wrong legend.
> I think this issue was fixed a while ago.
>
I'm pretty sure that's as it always has been. The string is fed
straight to gnuplot and you have to use '.' as decimal separator.
Allin.
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
12:07 a.m.
On Fri, 25 Sep 2009, [ISO-8859-1] H�lio Guilherme wrote:
Allin Cottrell wrote:
> On Thu, 24 Sep 2009, Talha Yalta wrote:
>
>> I have been testing the improvements for model windows under changing
>> subsampling options. Thanks to Allins recent work, this functionality
>> looks very nice and works smoothly in general. I also have some
>> suggestions and bugs to report:
[snip]
>
>> 4)- The subsampling info at the bottom of the gretl main window could
>> use some improvements. For example, when the sub sample is based on a
>> dummy, instead of just saying "current sample n = 100" it would be
>> nice if gretl said something like: "current sample based on dummy
>> <varname> (n = 100)" (The same thing for other subsampling options
>> such as random subsamples etc.)
>
> Too much information to print at the bottom of the window. I've
> added a menu item /Sample/Show status which shows some more detail
> than the bottom-of-the-screen line. It doesn't yet show a
> sampling criterion or anything of that sort -- that could maybe be
> added later (it would require recording more dataset information
> than we currently record).
>
Please do not forget to mark it for translation ;).
Thanks for the reminder! I'll do that.
>> 8)- My system is Turkish and when I use gretl in English,
the legend
>> for the OLS fitted line still uses commas as the decimal seperator.
>
> I'm not in a position to test that. Do other people see that too?
> (I.e. you're in a locale that uses the decimal comma, but you use
> the GUI selector to run in English: what decimal separator
> character do you see?)
>
I tested this and when in English the coeficients appear with decimal point.
pt_PT cvs version under Ubunto (Acer Aspire One :))
Thanks for testing this, H\'elio. Insofar as I can simulate the
situation Talha describes -- i.e. by setting the LANG and
LC_NUMERIC environnment variables to non-English values on my
computer then running gretl and choosing "English" via the GUI --
I get what you describe ('.' as the decimal character). So I'm
not sure what's going wrong in Talha's case.
Allin Cottrell
5:58 a.m.
On Thu, Sep 24, 2009 at 7:19 PM, Allin Cottrell <cottrell(a)wfu.edu> wrote:
On Thu, 24 Sep 2009, Talha Yalta wrote:
> 1)- In the Edit menu for the model window, it could be useful to
add
> an option such as "restore model sample" to quickly reverting to the
> subsample used by that model.
Could be nice, not very high priority.
This can be more useful than it seems at
first. Currently, gretl asks
whether the user wants to revert to the sample of the model only once
during when the model window is opened. If there are multiple model
windows open, the user finds himself constantly closing and reopening
windows just to switch between different subsample choices. Each model
window already knows its subsample so why not make switching more
accessible?
> 2)- I understand disabling forecasts but don't see why gretl
disables
> saving the residuals, fitted values etc. from a model just because the
> current subsample is different. I think it is perfectly OK to save the
> residuals based on the model subsample with the outside observations
> having missing values After all, this is what gretl does when the
> current subsample happens to match the model subsample.
This is problematic if the model was estimated on a subsample and
the dataset is also currently subsampled, but in a different way
from the model. But some possibilities may be worth exploring.
Again, each model
window already knows its subsample so I don't see
why this can be more problematic than what gretl can currently do,
that is running the various non-linearity, normality etc. tests
without the user being required to switch to the model sample. If I
understand correctly, quickly switching to the model sample and back
is what gretl is now doing in order to run these tests so the same
approach can be used to create a new variable where the sample
observations have the required data and the other observations are
just missing vaues.
> 3)- A Bug: I have 2 models one is with n=100 and based on a
dummy,
> while the other is using a random subsample with n=200. After
> estimating the second model, I return to the first one and gretl asks
> me if I want to restore the sample. I say yes and now the sample size
> is 60 instead of 100 (probably because gretl took the observations
> only from the active random subsample of n=200 instead of the full
> sample)...
Thanks for the report; fixed in CVS.
Thanks very much for the rapid fix.
> 4)- The subsampling info at the bottom of the gretl main window
could
> use some improvements. For example, when the sub sample is based on a
> dummy, instead of just saying "current sample n = 100" it would be
> nice if gretl said something like: "current sample based on dummy
> <varname> (n = 100)" (The same thing for other subsampling options
> such as random subsamples etc.)
Too much information to print at the bottom of the window. I've
added a menu item /Sample/Show status which shows some more detail
than the bottom-of-the-screen line. It doesn't yet show a
sampling criterion or anything of that sort -- that could maybe be
added later (it would require recording more dataset information
than we currently record).
IMHO, this information (with the additional dataset info
required) is
very useful for not just the current sample but for all the saved
models. Can this be added to individual model windows as well?
> 6)- Opening a new data set, all open windows are closed except
for the
> gnuplot graphs.
OK, that could cause trouble in some cases. The graph windows
should now close too.
Works nicely.
> 7)- In the model specification window, when the "Set as
default"
> option is checked, it becomes unchecked the next time (although the
> dependent variable is set correctly and as expected)
OK, cosmetic. Will be fixed at some point.
Again thanks. Cosmetics are more
important than what most people
think. Little things like this can result in bad first impressions
which is especially bad for a program like gretl where the users trust
their data for performing extremely complex operations.
> 8)- My system is Turkish and when I use gretl in English, the
legend
> for the OLS fitted line still uses commas as the decimal seperator.
I'm not in a position to test that. Do other people see that too?
(I.e. you're in a locale that uses the decimal comma, but you use
the GUI selector to run in English: what decimal separator
character do you see?)
This is interesting especially since Helio reports not
having the same
problem. Can this be due to the Linux distro? I cannot currently test
this on windows since I have the English Windows installed. I will let
you know when I find out more.
> 9)- Also, when using gretl in Turkish, entering formula for a new
line
> I need to use dot as the decimal seperator. If I use the comma sign in
> the formula, gretl draws a wrong line and a differently wrong legend.
> I think this issue was fixed a while ago.
I'm pretty sure that's as it always has been. The string is fed
straight to gnuplot and you have to use '.' as decimal separator.
I think
it is OK that we have to use dot as the decimal seperator at
all times but it is not OK that gretl produces erronious output when
someone accidentally uses commas. We need a check and an error message
here.
Again thanks very much for your attention and rapid fixing the various
bugs/issiues that I have mentioned.
Cheers
A. Talha Yalta
--
“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)
--
6:38 p.m.
On Fri, 11 Sep 2009, Talha Yalta wrote:
2)- It would be really neat if there is an option for the model
table that makes it possible to set the number of digits or
decimal places for the coefficients as well as se's and whether
t stats will be shown etc. For example, I have a table with 8
models which could be much smaller if I show 3 digits.
Fair enough. I've added a "Reformat" button to the menu bar for
the model table window which lets you set the number of
significant figures to print (as well as switch between std errors
and t-stats). Since it's now redundent, I've removed the
"Options" menu item from the right-click popup for the model table
icon.
Also related, the LaTeX output for the model table uses hardcoded
\footnotesize commands for each standard error value. If I need to
make the whole table smaller, these remain unchanged resulting in a
weird looking table...
I've made this a macro: the standard errors/t-stats are now
printed at \subsize. This is defined as \footnotesize by default
but you can change it in the TeX file if you want.
Allin.
5571
days inactive
5585
days old
16 comments
3 participants
participants (3)
-
Allin Cottrell -
Hélio Guilherme -
Talha Yalta