7:32 a.m.
Hi,
Two questions related to the "print" command.
1) Is the "range" option supposed to work jointly with the "by-obs"
option?
When by-obs is omitted the full range of series is printed out despite the
specification for range. For example:
? dataset sortby absr1std
? print absr1std --byobs --range=46:50
absr1std
ME 1.426741
MT 1.460884
VT 1.742409
DC 2.616447
FL 2.902663
? print absr1std --range=46:50
Full data range: 1 - 51 (n = 51)
0.0805403 0.0908925 0.0932311 0.0987927 0.127675 0.143004 0.164553
0.193521 0.203738 0.257765 0.280333 0.304542 0.307398 0.352839
... # I'm cutting here!!
? dataset sortby state
2) I'd like to print out the 5 greatest values of a series. As you may see
above, I first sort the whole dataset, then use print with range option and
finally sort again the dataset to revert to the initial position. Ok, these
operations come with no effort at all but, and this is just a suggestion,
would it be more intuitive if a sortby option be included in the print
command? Or a "dummy" option?
Best,
Artur
11:11 a.m.
On Wed, 23 Dec 2020, Artur Bala wrote:
2) I'd like to print out the 5 greatest values of a series. As
you may see
above, I first sort the whole dataset, then use print with range option and
finally sort again the dataset to revert to the initial position. Ok, these
operations come with no effort at all but, and this is just a suggestion,
would it be more intuitive if a sortby option be included in the print
command? Or a "dummy" option?
you can use the "ranking" function, as in
<hansl>
open greene9_1.gdt
smpl ranking(-labor) <= 5 --restrict
print labor --byobs
</hansl>
I'll admit it's not the most intuitive thing, but seems to work ok.
-------------------------------------------------------
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
-------------------------------------------------------
4:01 p.m.
Probably not very suggestive but pretty fair...and it works :)
Thank you Riccardo
Sent from my iPhone
> On Dec 23, 2020, at 12:11 PM, Riccardo (Jack) Lucchetti
<p002264(a)staff.univpm.it> wrote:
>
> On Wed, 23 Dec 2020, Artur Bala wrote:
>
>> 2) I'd like to print out the 5 greatest values of a series. As you may see
>> above, I first sort the whole dataset, then use print with range option and
>> finally sort again the dataset to revert to the initial position. Ok, these
>> operations come with no effort at all but, and this is just a suggestion,
>> would it be more intuitive if a sortby option be included in the print
>> command? Or a "dummy" option?
>
> you can use the "ranking" function, as in
>
> <hansl>
> open greene9_1.gdt
> smpl ranking(-labor) <= 5 --restrict
> print labor --byobs
> </hansl>
>
> I'll admit it's not the most intuitive thing, but seems to work ok.
>
>
> -------------------------------------------------------
> 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
>
-------------------------------------------------------_______________________________________________
> Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it
> To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it
> Website: https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/
4:03 p.m.
Oh, that’s good!!! Probably not very suggestive but pretty fair though :)
Thank you Riccardo
Sent from my iPhone
> On Dec 23, 2020, at 5:01 PM, Artur Bala <artur.bala.tn(a)gmail.com> wrote:
>
> Probably not very suggestive but pretty fair...and it works :)
> Thank you Riccardo
>
> Sent from my iPhone
>
>>> On Dec 23, 2020, at 12:11 PM, Riccardo (Jack) Lucchetti
<p002264(a)staff.univpm.it> wrote:
>>>
>>> On Wed, 23 Dec 2020, Artur Bala wrote:
>>>
>>> 2) I'd like to print out the 5 greatest values of a series. As you may
see
>>> above, I first sort the whole dataset, then use print with range option and
>>> finally sort again the dataset to revert to the initial position. Ok, these
>>> operations come with no effort at all but, and this is just a suggestion,
>>> would it be more intuitive if a sortby option be included in the print
>>> command? Or a "dummy" option?
>>
>> you can use the "ranking" function, as in
>>
>> <hansl>
>> open greene9_1.gdt
>> smpl ranking(-labor) <= 5 --restrict
>> print labor --byobs
>> </hansl>
>>
>> I'll admit it's not the most intuitive thing, but seems to work ok.
>>
>>
>> -------------------------------------------------------
>> 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
>>
-------------------------------------------------------_______________________________________________
>> Gretl-users mailing list -- gretl-users(a)gretlml.univpm.it
>> To unsubscribe send an email to gretl-users-leave(a)gretlml.univpm.it
>> Website: https://gretlml.univpm.it/postorius/lists/gretl-users.gretlml.univpm.it/
9:56 a.m.
On Wed, 23 Dec 2020, Artur Bala wrote:
Oh, that’s good!!! Probably not very suggestive but pretty fair
though :)
This one's a bit more sophisticated. Perhaps we could add this to the
"extra" package?
<hansl>
set verbose off
function void topbot(series x, scalar n[5])
genr index
catch markers --to-array=mrk
has_markers = !$error
smpl --no-missing
matrix tmp = {index, x}
r = rows(tmp)
tmp = msortby(tmp, 2)[,1]
selbot = tmp[1:n]
seltop = mreverse(tmp[r-n+1:r])
smpl full
printf "\nTOP BOTTOM\n"
printf "--------------------------------------------\n"
loop i = 1 .. n
kt = seltop[i]
kb = selbot[i]
if has_markers
printf "%10s: %10.3g", mrk[kt], x[kt]
printf "%10s: %10.3g\n", mrk[kb], x[kb]
else
printf "%10d: %10.3g", kt, x[kt]
printf "%10d: %10.3g\n", kb, x[kb]
endif
endloop
print ""
end function
# test without markers
open data4-1.gdt --quiet
topbot(price)
price[3] = NA # with missing values
topbot(price)
# test with markers
open greene9_1.gdt --quiet
topbot(labor,6)
labor[7] = NA # with missing values
topbot(labor,6)
</hansl>
-------------------------------------------------------
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
-------------------------------------------------------
2:03 p.m.
Am 24.12.2020 um 10:56 schrieb Riccardo (Jack) Lucchetti:
On Wed, 23 Dec 2020, Artur Bala wrote:
> Oh, that’s good!!! Probably not very suggestive but pretty fair
> though :)
This one's a bit more sophisticated. Perhaps we could add this to the
"extra" package?
I'm sorry, can you repeat again what's the problem that needs to be
solved by this code?
thanks
sven
4:05 p.m.
Le sam. 26 déc. 2020 à 15:03, Sven Schreiber <svetosch(a)gmx.net> a écrit :
Am 24.12.2020 um 10:56 schrieb Riccardo (Jack) Lucchetti:
> On Wed, 23 Dec 2020, Artur Bala wrote:
>
>> Oh, that’s good!!! Probably not very suggestive but pretty fair
>> though :)
>
> This one's a bit more sophisticated. Perhaps we could add this to the
> "extra" package?
>
I'm sorry, can you repeat again what's the problem that needs to be
solved by this code?
Hi Sven,
The initial problem was to print out "m" greatest values of a series in an
descending order.
Best
Artur
6:02 p.m.
Am 26.12.2020 um 17:05 schrieb Artur Bala:
Le sam. 26 déc. 2020 à 15:03, Sven Schreiber <svetosch(a)gmx.net
<mailto:svetosch@gmx.net>> a écrit :
Am 24.12.2020 um 10:56 schrieb Riccardo (Jack) Lucchetti:
> On Wed, 23 Dec 2020, Artur Bala wrote:
>
>> Oh, that’s good!!! Probably not very suggestive but pretty fair
>> though :)
>
> This one's a bit more sophisticated. Perhaps we could add this to the
> "extra" package?
>
I'm sorry, can you repeat again what's the problem that needs to be
solved by this code?
Hi Sven,
The initial problem was to print out "m" greatest values of a series in
an descending order.
Right, sorry, I thought I was missing something. In the GUI this is
actually trivial in the sense that you can even sort a time series there
in the "show values" window. But in the CLI of course one doesn't want
to apply "dataset sortby" to a time series!
Coming back to your original worry about print with the --range option,
indeed it looks like a bug to me. (I also observe that "print ...
--range=x:y" doesn't honor the x and y values if the option --byobs is
_absent_.)
cheers
sven
7:13 p.m.
On Sat, 26 Dec 2020, Sven Schreiber wrote:
Am 26.12.2020 um 17:05 schrieb Artur Bala:
> Le sam. 26 déc. 2020 à 15:03, Sven Schreiber <svetosch(a)gmx.net
> <mailto:svetosch@gmx.net>> a écrit :
>
> Am 24.12.2020 um 10:56 schrieb Riccardo (Jack) Lucchetti:
> > On Wed, 23 Dec 2020, Artur Bala wrote:
> >
> >> Oh, that’s good!!! Probably not very suggestive but pretty fair
> >> though :)
> >
> > This one's a bit more sophisticated. Perhaps we could add this to
> the
> > "extra" package?
> >
>
> I'm sorry, can you repeat again what's the problem that needs to be
> solved by this code?
>
>
> Hi Sven,
> The initial problem was to print out "m" greatest values of a series in
> an descending order.
Right, sorry, I thought I was missing something. In the GUI this is
actually trivial in the sense that you can even sort a time series there
in the "show values" window. But in the CLI of course one doesn't want
to apply "dataset sortby" to a time series!
No, but what about this? (Sorry if I'm missing the point or
repeating what has been said already.)
<hansl>
open AWM
# show biggest 5 values
eval dsort({GCR_YER})[1:5]
</hansl>
Coming back to your original worry about print with the --range
option,
indeed it looks like a bug to me. (I also observe that "print ...
--range=x:y" doesn't honor the x and y values if the option --byobs is
_absent_.)
OK, I suppose that should work without --byobs.
Allin
10:01 p.m.
Am 26.12.2020 um 20:13 schrieb Allin Cottrell:
On Sat, 26 Dec 2020, Sven Schreiber wrote:
>
> Right, sorry, I thought I was missing something. In the GUI this is
> actually trivial in the sense that you can even sort a time series there
> in the "show values" window. But in the CLI of course one doesn't want
> to apply "dataset sortby" to a time series!
No, but what about this? (Sorry if I'm missing the point or repeating
what has been said already.)
<hansl>
open AWM
# show biggest 5 values
eval dsort({GCR_YER})[1:5]
</hansl>
I think the observation numbers or markers were also wanted. Perhaps
like this:
<hansl>
open AWM
# show biggest 5 values
series date = $obsdate
list L = GCR_YER date
eval mreverse(msortby({L}, 1)[1:5,])
</hansl>
But the number format of the date is not adequate there.
cheers
sven
10:15 p.m.
On Sat, 26 Dec 2020, Sven Schreiber wrote:
Am 26.12.2020 um 20:13 schrieb Allin Cottrell:
> On Sat, 26 Dec 2020, Sven Schreiber wrote:
>>
>> Right, sorry, I thought I was missing something. In the GUI this is
>> actually trivial in the sense that you can even sort a time series there
>> in the "show values" window. But in the CLI of course one doesn't
want
>> to apply "dataset sortby" to a time series!
>
> No, but what about this? (Sorry if I'm missing the point or repeating
> what has been said already.)
>
> <hansl>
> open AWM
> # show biggest 5 values
> eval dsort({GCR_YER})[1:5]
> </hansl>
I think the observation numbers or markers were also wanted. Perhaps
like this:
<hansl>
open AWM
# show biggest 5 values
series date = $obsdate
list L = GCR_YER date
eval mreverse(msortby({L}, 1)[1:5,])
</hansl>
But the number format of the date is not adequate there.
So:
<hansl>
open AWM
m = mreverse(msortby({obs, GCR_YER},2))[1:5,]
strings S = array(5)
loop i=1..5
S[i] = obslabel(m[i,1])
endloop
rnameset(m, S)
print m
</hansl>
But by now it's not quite so trivial.
Allin
12:55 a.m.
On Sat, 26 Dec 2020, Allin Cottrell wrote:
On Sat, 26 Dec 2020, Sven Schreiber wrote:
>
> I think the observation numbers or markers were also wanted. Perhaps
> like this:
> <hansl>
> open AWM
> # show biggest 5 values
> series date = $obsdate
> list L = GCR_YER date
> eval mreverse(msortby({L}, 1)[1:5,])
> </hansl>
>
> But the number format of the date is not adequate there.
So:
<hansl>
open AWM
m = mreverse(msortby({obs, GCR_YER},2))[1:5,]
strings S = array(5)
loop i=1..5
S[i] = obslabel(m[i,1])
endloop
rnameset(m, S)
print m
</hansl>
New in git, obslabel() supports a vector argument, so this is a bit
less hassle:
<hansl>
open AWM
m = mreverse(msortby({obs, GCR_YER},2))[1:5,]
rnameset(m, obslabel(m[,1]))
printf "%10g\n", m
</hansl>
Allin
10:58 a.m.
Am 27.12.2020 um 01:55 schrieb Allin Cottrell:
New in git, obslabel() supports a vector argument, so this is a bit
less
hassle:
Thanks, that's useful.
<hansl>
open AWM
m = mreverse(msortby({obs, GCR_YER},2))[1:5,]
rnameset(m, obslabel(m[,1]))
printf "%10g\n", m
</hansl>
BTW, I only found a mention of the "obs" special variable in the
cheatsheet chapter of the guide. And it is used in the doc for the
"smpl" command, but arguably it's not clear there that it can be used
more generally - also it does not appear in the subsampling chapter 4 in
the guide. (However, it does appear inside the --declared as somewhat
outdated-- genr entry of the command reference.)
So, some questions:
Might it be an idea to add a hint about "obs" to the series section 11.2
of the guide? Also, apart from historical reasons, why isn't this thing
actually called $obs, like $obsdate and so on? I also seem to remember
"t" working in the same way - what's the status there?
thanks
sven
3:36 p.m.
On Sat, 26 Dec 2020, Sven Schreiber wrote:
Coming back to your original worry about print with the --range
option,
indeed it looks like a bug to me. (I also observe that "print ...
--range=x:y" doesn't honor the x and y values if the option --byobs is
_absent_.)
That's now fixed.
Allin
1459
days inactive
1464
days old
13 comments
4 participants
participants (4)
-
Allin Cottrell -
Artur Bala -
Riccardo (Jack) Lucchetti -
Sven Schreiber