4:28 p.m.
Hello all,
I attach two gdt panel data files (is what I managed to do, TIME is a time
index, GEO lists 38 countries/regions, and two variables CP00, CP01),
example1.gdt goes from 2016:01-2018:12 and
example2.gdt goes from 2019:01-2022:04
(A) is there an "easy" way to "join" the two gdt files into one
time-larger
file? running from 2016:01:-2022:04?
(B) how can i "easily" transform the panel dataset into a time-series set
with 38 CP00 series, say EU_CP00, EA_CP00,... and 38 CP01 series, EU_CP01,
EA_CP01,...?
PS = OK I admit "easy" is rather varue. I mean using the mouse menu route
or i guess a few hansl lines? (of course many more lines would be
also welcome if there is no other way ...).
Thanks
Yiannis
5:23 p.m.
Am 07.06.22 um 18:28 schrieb Ioannis Venetis:
Hello all,
I attach two gdt panel data files (is what I managed to do, TIME is a
time index, GEO lists 38 countries/regions, and two variables CP00, CP01),
example1.gdt goes from 2016:01-2018:12 and
example2.gdt goes from 2019:01-2022:04
(A) is there an "easy" way to "join" the two gdt files into one
time-larger file? running from 2016:01:-2022:04?
(B) how can i "easily" transform the panel dataset into a time-series
set with 38 CP00 series, say EU_CP00, EA_CP00,... and 38 CP01
series, EU_CP01, EA_CP01,...?
PS = OK I admit "easy" is rather varue. I mean using the mouse menu
route or i guess a few hansl lines? (of course many more lines would be
also welcome if there is no other way ...).
Hi,
joining datasets is easy using either the GUI or hansl. Please have a
look at ch. 7 in the user-guide which is quite detailed on that topic.
Artur
10:01 p.m.
Am 07.06.2022 um 18:28 schrieb Ioannis Venetis:
Hello all,
Hi Ioannis,
I attach two gdt panel data files (is what I managed to do, TIME is a
time index, GEO lists 38 countries/regions, and two variables CP00, CP01),
example1.gdt goes from 2016:01-2018:12 and
example2.gdt goes from 2019:01-2022:04
Actually it seems to be the other way around.
(A) is there an "easy" way to "join" the two gdt
files into one
time-larger file? running from 2016:01:-2022:04?
I think the trick here would be to first extend the time dimension of
the earlier dataset (example2) to cover everything up to 2022, being
empty until the actual appending is done. (menu Data/Add observations)
Simple appending still doesn't seem to work then (maybe it should), but
you can use the graphical join option, then in the dialog window specify
"TIME" and "GEO" as the keys to be matched. In the GUI right now you
have to do that separately for each variable, I'm afraid. (In a script
it's easier for multiple variables, but here you only have two.)
Hope that explanation was clear.
(B) how can i "easily" transform the panel dataset into a
time-series
set with 38 CP00 series, say EU_CP00, EA_CP00,... and 38 CP01
series, EU_CP01, EA_CP01,...?
Ah, that's a classic - well, I just mean that I've had this type of
problem myself. I don't think there's a super easy way. One solution is
to convert the stuff to a matrix, reshape that, and then reload it into
a clean time series dataset. The key is to know that gretl's panel
storage is as stacked time series, so a series becomes an NT,1 vector
(when missings are included). Something like:
<hansl>
set skip_missing off
matrix M00 = {CP00}
N = max($unit)
T = max($obsminor) # $obsminor is time in panels
M00 = mshape(M00, T, N)
</hansl>
And then you would perhaps store the matrix M00, prepare a time series
dataset of length T, and load M00 and turn it into your series
collection. (Perhaps using the mat2list function.)
HTH,
sven
8:58 a.m.
On Tue, 7 Jun 2022, Sven Schreiber wrote:
Am 07.06.2022 um 18:28 schrieb Ioannis Venetis:
> Hello all,
Hi Ioannis,
>
> I attach two gdt panel data files (is what I managed to do, TIME is a
> time index, GEO lists 38 countries/regions, and two variables CP00, CP01),
>
> example1.gdt goes from 2016:01-2018:12 and
> example2.gdt goes from 2019:01-2022:04
Actually it seems to be the other way around.
> (A) is there an "easy" way to "join" the two gdt files into one
> time-larger file? running from 2016:01:-2022:04?
I think the trick here would be to first extend the time dimension of
the earlier dataset (example2) to cover everything up to 2022, being
empty until the actual appending is done. (menu Data/Add observations)
Simple appending still doesn't seem to work then (maybe it should), but
you can use the graphical join option, then in the dialog window specify
"TIME" and "GEO" as the keys to be matched. In the GUI right now you
have to do that separately for each variable, I'm afraid. (In a script
it's easier for multiple variables, but here you only have two.)
Hope that explanation was clear.
Simple appending does work if you get rid of the panel info and then
reinstate it, as in
<hansl>
open example1.gdt
setobs 1 1
store flat1.gdt
open example2.gdt
setobs 1 1
append flat1.gdt
setobs GEO TIME --panel-vars
</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
-------------------------------------------------------
10:22 a.m.
Am 08.06.2022 um 10:58 schrieb Riccardo (Jack) Lucchetti:
On Tue, 7 Jun 2022, Sven Schreiber wrote:
> Simple appending still doesn't seem to work then (maybe it should), but
> you can use the graphical join option, then in the dialog window specify
> "TIME" and "GEO" as the keys to be matched. In the GUI right now
you
> have to do that separately for each variable, I'm afraid. (In a script
> it's easier for multiple variables, but here you only have two.)
>
> Hope that explanation was clear.
Simple appending does work if you get rid of the panel info and then
reinstate it, as in
<hansl>
open example1.gdt
setobs 1 1
store flat1.gdt
open example2.gdt
setobs 1 1
append flat1.gdt
setobs GEO TIME --panel-vars
</hansl>
Very nice workaround! IMHO this, however, shows that gretl is still not
able to treat its own panel datasets as first-class citizens. (Although
much progress has been made recently, no doubt!)
cheers
sven
11:37 a.m.
On Tue, Jun 7, 2022 at 12:29 PM Ioannis Venetis
<ioannisvenetis(a)gmail.com> wrote:
I attach two gdt panel data files (is what I managed to do, TIME is a time index, GEO
lists 38 countries/regions, and two variables CP00, CP01),
example1.gdt goes from 2016:01-2018:12 and
example2.gdt goes from 2019:01-2022:04
(A) is there an "easy" way to "join" the two gdt files into one
time-larger file? running from 2016:01:-2022:04?
(B) how can i "easily" transform the panel dataset into a time-series set with
38 CP00 series, say EU_CP00, EA_CP00,... and 38 CP01 series, EU_CP01, EA_CP01,...?
On question (B) here's a slight refinement of Sven's suggestion:
<hansl>
open example1.gdt
set skip_missing off
matrix X = mshape({CP00}, $pd)
S = strvals(GEO)
loop i=1..nelem(S)
S[i] = "CP00_" ~ S[i]
endloop
cnameset(X, S)
T = rows(X)
nulldata T --preserve
setobs 12 2019:01
list CP00 = mat2list(X)
print CP00 -o
</hansl>
Allin
1:28 p.m.
On Wed, 8 Jun 2022, Cottrell, Allin wrote:
On Tue, Jun 7, 2022 at 12:29 PM Ioannis Venetis
<ioannisvenetis(a)gmail.com> wrote:
>
> I attach two gdt panel data files (is what I managed to do, TIME is a time index, GEO
lists 38 countries/regions, and two variables CP00, CP01),
>
> example1.gdt goes from 2016:01-2018:12 and
> example2.gdt goes from 2019:01-2022:04
>
> (A) is there an "easy" way to "join" the two gdt files into one
time-larger file? running from 2016:01:-2022:04?
> (B) how can i "easily" transform the panel dataset into a time-series set
with 38 CP00 series, say EU_CP00, EA_CP00,... and 38 CP01 series, EU_CP01, EA_CP01,...?
On question (B) here's a slight refinement of Sven's suggestion:
<hansl>
open example1.gdt
set skip_missing off
matrix X = mshape({CP00}, $pd)
S = strvals(GEO)
loop i=1..nelem(S)
S[i] = "CP00_" ~ S[i]
endloop
cnameset(X, S)
T = rows(X)
nulldata T --preserve
setobs 12 2019:01
list CP00 = mat2list(X)
print CP00 -o
</hansl>
In the spirit of "it's nice to have more ways to do something", here's
an
alternative (probably inferior) solution.
<hansl>
clear
open example1.gdt
S = strvals(GEO)
T = $pd
nulldata T --preserve
setobs 12 2019:01
loop i=1..nelem(S)
c = S[i]
flt = sprintf("GEO==\\\"%s\\\"", c)
fn = fixname(c)
join example1.gdt @fn --data="CP00" --filter="@flt"
endloop
</hansl>
That said, it may be nice to write a function package providing a function
like this, with a signature like
function scalar unstack(list X, string fname)
which would save the time-series data into "fname.gdt" (I imagine it would
return an error code).
-------------------------------------------------------
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
-------------------------------------------------------
686
days inactive
687
days old
6 comments
5 participants
participants (5)
-
Artur T. -
Cottrell, Allin -
Ioannis Venetis -
Riccardo (Jack) Lucchetti -
Sven Schreiber