9:23 p.m.
Hello All,
I am ex-Stata user trying to work my way into gretl - though slowly. I have a panel data
set in Excel for 56 state counties (identified by a unique county code 01 - 56) over 51
years (annual data 1990 - 2040). I seem to be able to load the data set into gretl and
get it to recognize the panel nature of the data, but I can't seem to get it to
acknowledge the labels for those years (i.e. 1990 - 2040) when using commands such as
gnuplot for graphing.
For example, when I load the data into gretl, it recognizes the panel as "Panel: Full
range 1:01 - 56:51". When I restrict the full data set to say two counties and graph
them (using "gnuplot County1 County2 -time-series -with-lines") the data are
plotted on a graph with an x-axis scale from 1-51 instead of 1990 - 2040. How can I get
it to recognize the data as that will create x-axis labels in -time-series graphs
automatically with the respective dates rather than the periods 1 - 51?
Thank you for your time on this!
Joe
3:26 p.m.
On Tue, 19 Feb 2013, Ramler, Joe wrote:
I am ex-Stata user trying to work my way into gretl - though
slowly.
Welcome to gretl-land!
With regard to panel data...
For example, when I load the data into gretl, it recognizes
the panel as "Panel: Full range 1:01 - 56:51". When I
restrict the full data set to say two counties and graph
them (using "gnuplot County1 County2 -time-series
-with-lines") the data are plotted on a graph with an x-axis
scale from 1-51 instead of 1990 - 2040. How can I get it to
recognize the data as that will create x-axis labels in
-time-series graphs automatically with the respective dates
rather than the periods 1 - 51?
Right now there's no automatic way to do that. If a dataset
contains a "year" variable (say), that can be used in setting
the panel structure but there's no way of telling gretl, "I
want this to be the time variable for the panel, please use it
in graphing."
However, this would be a nice thing to have. We could add
another member to the set of state variables that can be
manipulated by the "set" command, something like
set panel_time year
where "panel_time" is the name of the "set" variable and
"year" is the name of a series in the dataset.
I'll experiment with this.
Allin Cottrell
11:24 p.m.
On Tue, 19 Feb 2013, Ramler, Joe wrote:
For example, when I load the data into gretl, it recognizes the
panel as "Panel: Full range 1:01 - 56:51". When I restrict the
full data set to say two counties and graph them (using "gnuplot
County1 County2 -time-series -with-lines") the data are plotted on
a graph with an x-axis scale from 1-51 instead of 1990 - 2040.
How can I get it to recognize the data as that will create x-axis
labels in -time-series graphs automatically with the respective
dates rather than the periods 1 - 51?
I should have asked: have you tried playing with "Panel Plot" in the
GUI? It's available via the right-click menu when you select a
single series in the main window. If there's a variable named "year"
or "period" in the dataset (on a case-insensitive comparison), and
it appears to be suitable to the purpose, it is used for the x-axis
on the time-series plot options "single graph: group means",
"single graph: groups overlaid" and "multiple plots arranged
vertically".
Allin Cottrell
4:20 p.m.
Excuse my ignorance, but if you're talking about the drop-down features, I did try
those. The issue I'm having with those is I'd like to be able to script all of
this in an attempt to create an automated "report creator", if you will, using
LaTeX. This data is going to be updated fairly frequently, and I don't really want to
have to run 100+ graphs by hand (using the drop down features) every time the data
changes; having a boiler plate script that could be run would be ideal. Thanks for your
time on this, Allin. I really do appreciate it.
Best,
Joe
-----Original Message-----
From: gretl-users-bounces(a)lists.wfu.edu [mailto:gretl-users-bounces@lists.wfu.edu] On
Behalf Of Allin Cottrell
Sent: Wednesday, February 20, 2013 4:24 PM
To: Gretl list
Subject: Re: [Gretl-users] Defining Panel Data
On Tue, 19 Feb 2013, Ramler, Joe wrote:
For example, when I load the data into gretl, it recognizes the panel
as "Panel: Full range 1:01 - 56:51". When I restrict the full data
set to say two counties and graph them (using "gnuplot
County1 County2 -time-series -with-lines") the data are plotted on a
graph with an x-axis scale from 1-51 instead of 1990 - 2040.
How can I get it to recognize the data as that will create x-axis
labels in -time-series graphs automatically with the respective dates
rather than the periods 1 - 51?
I should have asked: have you tried playing with "Panel Plot" in the GUI?
It's available via the right-click menu when you select a single series in the main
window. If there's a variable named "year"
or "period" in the dataset (on a case-insensitive comparison), and it appears to
be suitable to the purpose, it is used for the x-axis on the time-series plot options
"single graph: group means", "single graph: groups overlaid" and
"multiple plots arranged vertically".
Allin Cottrell
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
5:06 p.m.
On Thu, 21 Feb 2013, Ramler, Joe wrote:
Excuse my ignorance, but if you're talking about the drop-down
features,
I did try those. The issue I'm having with those is I'd like to be able
to script all of this in an attempt to create an automated "report
creator", if you will, using LaTeX. This data is going to be updated
fairly frequently, and I don't really want to have to run 100+ graphs by
hand (using the drop down features) every time the data changes; having
a boiler plate script that could be run would be ideal. Thanks for your
time on this, Allin. I really do appreciate it.
Is this close to what you need?
<hansl>
# example fake data
nulldata 2856
series year = 1990 + (time-1)%51
series county = 1 + floor((time-1)/51)
# set up panel structure
setobs county year --panel-vars
# generate artificial AR(2) series, just for fun
series x = 0
series x = 1.4 * x(-1) - 0.5*x(-2) + normal()
loop foreach i 23 34 45 --quiet
smpl county=$i --restrict --replace
string fname="cnty_$i.pdf"
gnuplot x year --with-lines --output=@fname --suppress-fitted
end loop
smpl full
</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
-------------------------------------------------------
4338
days inactive
4340
days old
4 comments
3 participants
participants (3)
-
Allin Cottrell -
Ramler, Joe -
Riccardo (Jack) Lucchetti