5:19 p.m.
Dear Gretl-Devs,
when "printing" summary statistics, correlation matrixes, etc., the usual limit
of 31 characters for variables is not held. Instead, variables are truncated to 15
characters... In some cases making them indistinguishable. If it was possible to also use
the limit of 31 chars here, that would be really nice.
Kind Regards
Frederik Schaff
------------
Dr. Frederik Schaff
FernUniversität in Hagen
Department of Business and Economics
Chair of Economic Theory (Prof. A. Endres)
Universitätsstraße 11 (TGZ)
58097 Hagen
Voice: +49 (0) 2331 987-4454
E-mail: Frederik.Schaff@FernUni-Hagen.de<mailto:Frederik.Schaff@FernUni-Hagen.de>
WWW: e.feu.de/schaff
5:23 p.m.
Not sure that should be realized. The output must look awful with more
than 15 characters, right? ;-) What about shorter variable names, Frederik?
Best,
Artur
Am 28.02.2017 um 18:19 schrieb Schaff, Frederik:
Dear Gretl-Devs,
when „printing“ summary statistics, correlation matrixes, etc., the
usual limit of 31 characters for variables is not held. Instead,
variables are truncated to 15 characters... In some cases making them
indistinguishable. If it was possible to also use the limit of 31 chars
here, that would be really nice.
Kind Regards
Frederik Schaff
5:31 p.m.
Ah well, that is sometimes not practical for me ;)
Consider the following: I have a variable called "MA_Utility" which is the
moving average utility of a single agent. Then I have some statistics I take during the
run to analyse the distribution later. E.g. the MAE and, because it is a distribution at
each point in time and in time itself, I have e.g. the MAE of the MAE of the MA_Utility,
the RMSE of the AVG of the MA_Utility, the Avg of the Min of the MA_Utility etc...
Now, this looks ugly:
MA_Utility_MAE~ MA_Utility_MAE~ MA_Utility_MAE~
1.0000 0.8500 0.1773 MA_Utility_MAE~
1.0000 0.4378 MA_Utility_MAE~
1.0000 MA_Utility_MAE~
;) When publishing some such stuff, I will use other kinds of notation so it looks not as
ugly, of course (like mathematical notation, e.g.).
Regards,
Frederik
-----Ursprüngliche Nachricht-----
Von: gretl-devel-bounces(a)lists.wfu.edu [mailto:gretl-devel-
bounces(a)lists.wfu.edu] Im Auftrag von Artur Tarassow
Gesendet: Dienstag, 28. Februar 2017 18:24
An: gretl-devel(a)lists.wfu.edu
Betreff: Re: [Gretl-devel] Variable names in summary statistics (and similar)
Not sure that should be realized. The output must look awful with more than
15 characters, right? ;-) What about shorter variable names, Frederik?
Best,
Artur
Am 28.02.2017 um 18:19 schrieb Schaff, Frederik:
> Dear Gretl-Devs,
>
>
>
> when "printing" summary statistics, correlation matrixes, etc., the
> usual limit of 31 characters for variables is not held. Instead,
> variables are truncated to 15 characters... In some cases making them
> indistinguishable. If it was possible to also use the limit of 31
> chars here, that would be really nice.
>
>
>
> Kind Regards
>
> Frederik Schaff
>
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
5:37 p.m.
Of course I understand your point. But the example may not be very
convincing: just replace "Utility" by "U" and you save 6 characters
;-)
But let's see what others say.
Good luck,
Artur
Am 28.02.2017 um 18:31 schrieb Schaff, Frederik:
Ah well, that is sometimes not practical for me ;)
Consider the following: I have a variable called "MA_Utility" which is the
moving average utility of a single agent. Then I have some statistics I take during the
run to analyse the distribution later. E.g. the MAE and, because it is a distribution at
each point in time and in time itself, I have e.g. the MAE of the MAE of the MA_Utility,
the RMSE of the AVG of the MA_Utility, the Avg of the Min of the MA_Utility etc...
Now, this looks ugly:
MA_Utility_MAE~ MA_Utility_MAE~ MA_Utility_MAE~
1.0000 0.8500 0.1773 MA_Utility_MAE~
1.0000 0.4378 MA_Utility_MAE~
1.0000 MA_Utility_MAE~
;) When publishing some such stuff, I will use other kinds of notation so it looks not as
ugly, of course (like mathematical notation, e.g.).
Regards,
Frederik
>
> Not sure that should be realized. The output must look awful with more than
> 15 characters, right? ;-) What about shorter variable names, Frederik?
>
> Best,
> Artur
>
> Am 28.02.2017 um 18:19 schrieb Schaff, Frederik:
>> Dear Gretl-Devs,
>>
>>
>>
>> when "printing" summary statistics, correlation matrixes, etc., the
>> usual limit of 31 characters for variables is not held. Instead,
>> variables are truncated to 15 characters... In some cases making them
>> indistinguishable. If it was possible to also use the limit of 31
>> chars here, that would be really nice.
>>
>>
>>
>> Kind Regards
>>
>> Frederik Schaff
>>
5:45 p.m.
On Tue, 28 Feb 2017, Artur Tarassow wrote:
Of course I understand your point. But the example may not be very
convincing: just replace "Utility" by "U" and you save 6 characters
;-) But
let's see what others say.
I think that if you want to print a correlation matrix using very long
variable names in full, you really have to design your own function
for that. The tools are available in hansl. The compact format used by
default by gretl is not going to work.
Allin
Am 28.02.2017 um 18:31 schrieb Schaff, Frederik:
> Ah well, that is sometimes not practical for me ;)
>
> Consider the following: I have a variable called "MA_Utility" which is the
> moving average utility of a single agent. Then I have some statistics I
> take during the run to analyse the distribution later. E.g. the MAE and,
> because it is a distribution at each point in time and in time itself, I
> have e.g. the MAE of the MAE of the MA_Utility, the RMSE of the AVG of the
> MA_Utility, the Avg of the Min of the MA_Utility etc...
>
>
> Now, this looks ugly:
>
> MA_Utility_MAE~ MA_Utility_MAE~ MA_Utility_MAE~
> 1.0000 0.8500 0.1773 MA_Utility_MAE~
> 1.0000 0.4378 MA_Utility_MAE~
> 1.0000 MA_Utility_MAE~
9:29 p.m.
Thanks for the answer. I thought it might be that one just needed to adjust, say, a
parameter, to make it consistent with the regular maximal variable names length but I
guess this would spoil the fixed layout. Another suggestion would be to somehow
"switch" to variables IDs if the name is too long. Something like ID_xyz or
such, instead of simply truncation with ~. Then I could use copy & paste to adjust the
names in the latex tables, without risking to mix everything up.
Best
Frederik
-----Ursprüngliche Nachricht-----
Von: gretl-devel-bounces(a)lists.wfu.edu [mailto:gretl-devel-
bounces(a)lists.wfu.edu] Im Auftrag von Allin Cottrell
Gesendet: Dienstag, 28. Februar 2017 18:46
An: Gretl development <gretl-devel(a)lists.wfu.edu>
Betreff: Re: [Gretl-devel] Variable names in summary statistics (and similar)
On Tue, 28 Feb 2017, Artur Tarassow wrote:
> Of course I understand your point. But the example may not be very
> convincing: just replace "Utility" by "U" and you save 6
characters
> ;-) But let's see what others say.
I think that if you want to print a correlation matrix using very long variable
names in full, you really have to design your own function for that. The tools
are available in hansl. The compact format used by default by gretl is not
going to work.
Allin
> Am 28.02.2017 um 18:31 schrieb Schaff, Frederik:
>> Ah well, that is sometimes not practical for me ;)
>>
>> Consider the following: I have a variable called "MA_Utility" which
>> is the moving average utility of a single agent. Then I have some
>> statistics I take during the run to analyse the distribution later.
>> E.g. the MAE and, because it is a distribution at each point in time
>> and in time itself, I have e.g. the MAE of the MAE of the MA_Utility,
>> the RMSE of the AVG of the MA_Utility, the Avg of the Min of the
MA_Utility etc...
>>
>>
>> Now, this looks ugly:
>>
>> MA_Utility_MAE~ MA_Utility_MAE~ MA_Utility_MAE~
>> 1.0000 0.8500 0.1773 MA_Utility_MAE~
>> 1.0000 0.4378 MA_Utility_MAE~
>> 1.0000 MA_Utility_MAE~
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
9:45 p.m.
On Tue, 28 Feb 2017, Schaff, Frederik wrote:
Thanks for the answer. I thought it might be that one just needed
to adjust, say, a parameter, to make it consistent with the
regular maximal variable names length but I guess this would spoil
the fixed layout. Another suggestion would be to somehow "switch"
to variables IDs if the name is too long. Something like ID_xyz or
such, instead of simply truncation with ~.
OK, that's an idea worth thinking about.
Then I could use copy & paste to adjust the names in the latex
tables, without risking to mix everything up.
My first thought is that with looong names you might want to depart
from the triangular matrix format altogether.
Allin
9:54 p.m.
Von: gretl-devel-bounces(a)lists.wfu.edu [mailto:gretl-devel-
bounces(a)lists.wfu.edu] Im Auftrag von Allin Cottrell
Gesendet: Dienstag, 28. Februar 2017 22:46
An: Gretl development <gretl-devel(a)lists.wfu.edu>
Betreff: Re: [Gretl-devel] Variable names in summary statistics (and similar)
On Tue, 28 Feb 2017, Schaff, Frederik wrote:
> Thanks for the answer. I thought it might be that one just needed to
> adjust, say, a parameter, to make it consistent with the regular
> maximal variable names length but I guess this would spoil the fixed
> layout. Another suggestion would be to somehow "switch"
> to variables IDs if the name is too long. Something like ID_xyz or
> such, instead of simply truncation with ~.
OK, that's an idea worth thinking about.
> Then I could use copy & paste to adjust the names in the latex tables,
> without risking to mix everything up.
My first thought is that with looong names you might want to depart from
the triangular matrix format altogether.
Allin
The ones I am actually using for publication or presentation I am past-processing anyway
(2 column names, as an example). ATM I cannot (without room for error) use the latex
export (or raw export into, say excel, where I can use Exel2LaTex to prepare the data
further for latex and fine tune it there later) because all the Information on the
variables is lost due to the truncation.
Frederik
10:14 p.m.
On Tue, 28 Feb 2017, Schaff, Frederik wrote:
> On Tue, 28 Feb 2017, Schaff, Frederik wrote:
>
>> Thanks for the answer. I thought it might be that one just needed to
>> adjust, say, a parameter, to make it consistent with the regular
>> maximal variable names length but I guess this would spoil the fixed
>> layout. Another suggestion would be to somehow "switch"
>> to variables IDs if the name is too long. Something like ID_xyz or
>> such, instead of simply truncation with ~.
>
> OK, that's an idea worth thinking about.
>
>> Then I could use copy & paste to adjust the names in the latex tables,
>> without risking to mix everything up.
>
> My first thought is that with looong names you might want to depart from
> the triangular matrix format altogether.
>
> Allin
The ones I am actually using for publication or presentation I am
past-processing anyway (2 column names, as an example). ATM I
cannot (without room for error) use the latex export (or raw
export into, say excel, where I can use Exel2LaTex to prepare the
data further for latex and fine tune it there later) because all
the Information on the variables is lost due to the truncation.
In that case I think Artur's suggestion is the right one: radically
shorten the variable names in some systematic fashion, then
reconstruct them algorithmically in your preferred final format. So
for example:
series MAU = MA_Utility
and also shorten the suffix tags:
MA_Utility_MAE -> MAU001
allowing you to encode 999 tags in a 6-character name ;-)
Allin
7:58 a.m.
In that case I think Artur's suggestion is the right one:
radically shorten the
variable names in some systematic fashion, then reconstruct them
algorithmically in your preferred final format. So for example:
series MAU = MA_Utility
and also shorten the suffix tags:
MA_Utility_MAE -> MAU001
allowing you to encode 999 tags in a 6-character name ;-)
Allin
Although there is a clear mapping than, I need to check tables all the while ... but it is
workable. Is there a way to "rename" a series rather than to copy it? And I
would need to automate it somehow ... I have to think about how to do so. Is there a way
to string-substitute parts or do you have an idea how to do so in gretl/hansl? Like
string_replace(Input,"MA_Utility","MAU") #in Input replace each
occurrence of "MA_Utility" with "U", if any.
so I could use it in a loop?
Far more simple would be the truncation of too long names to ID_xyz however ;)
Many thanks
8:09 a.m.
On Wed, 1 Mar 2017, Schaff, Frederik wrote:
Although there is a clear mapping than, I need to check tables all
the
while ... but it is workable. Is there a way to "rename" a series rather
than to copy it?
How about the "rename" command? ;)
-------------------------------------------------------
Riccardo (Jack) Lucchetti
Dipartimento di Scienze Economiche e Sociali (DiSES)
Università Politecnica delle Marche
(formerly known as Università di Ancona)
r.lucchetti(a)univpm.it
http://www2.econ.univpm.it/servizi/hpp/lucchetti
-------------------------------------------------------
8:25 a.m.
How about the "rename" command? ;)
That I should have seen by myself... thanks! And there I see "strsub",
which shall do the job if I can get the name of a variable...
Ok, so the following would be nice:
list alls = dataset
loop foreach i alls
string new = strsub(varname($i), "Utility", "U")
rename $i new
endloop
but it does not work, because varname (in my version) still needs an ID. When will the
recent updated that fixes this be pushed? I am using windows...
Many thanks!
8:29 a.m.
A couple of days ago this was changed:
"Now varname() accepts either an ID number or a series itself."
Are you using the latest cvs/git version?
Artur
Am 01.03.2017 um 09:25 schrieb Schaff, Frederik:
> How about the "rename" command? ;)
>
>
That I should have seen by myself... thanks! And there I see "strsub", which
shall do the job if I can get the name of a variable...
Ok, so the following would be nice:
list alls = dataset
loop foreach i alls
string new = strsub(varname($i), "Utility", "U")
rename $i new
endloop
but it does not work, because varname (in my version) still needs an ID. When will the
recent updated that fixes this be pushed? I am using windows...
Many thanks!
_______________________________________________
Gretl-devel mailing list
Gretl-devel(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-devel
8:48 a.m.
Sry again, I got it from the homepage (sourceforge) by selecting the "last
snapshot" and now it works (I used the updating prior...).
Here is the working example for the copy:
___
clear
nulldata 100
setobs 1 1 --cross-section
series somethingelse = uniform()
series Utility_min = normal()
series Utility_max = normal()+1
series Utility_avg = (Utility_min + Utility_max)/2
list alls = dataset
loop foreach i alls
string new = strsub(varname($i), "Utility", "U")
catch rename $i @new
endloop
___
Many thanks to all of you!
-----Ursprüngliche Nachricht-----
Von: gretl-devel-bounces(a)lists.wfu.edu [mailto:gretl-devel-
bounces(a)lists.wfu.edu] Im Auftrag von Artur Tarassow
Gesendet: Mittwoch, 1. März 2017 09:29
An: Gretl development <gretl-devel(a)lists.wfu.edu>
Betreff: Re: [Gretl-devel] Variable names in summary statistics (and similar)
A couple of days ago this was changed:
"Now varname() accepts either an ID number or a series itself."
Are you using the latest cvs/git version?
Artur
Am 01.03.2017 um 09:25 schrieb Schaff, Frederik:
>> How about the "rename" command? ;)
>>
>>
> That I should have seen by myself... thanks! And there I see "strsub",
which shall do the job if I can get the name of a variable...
>
> Ok, so the following would be nice:
>
> list alls = dataset
> loop foreach i alls
> string new = strsub(varname($i), "Utility", "U")
> rename $i new
> endloop
>
> but it does not work, because varname (in my version) still needs an ID.
When will the recent updated that fixes this be pushed? I am using
windows...
>
> Many thanks!
>
> _______________________________________________
> 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
2852
days inactive
2853
days old
13 comments
4 participants
participants (4)
-
Allin Cottrell -
Artur Tarassow -
Riccardo (Jack) Lucchetti -
Schaff, Frederik