11:48 a.m.
Am 25.05.2019 um 17:46 schrieb "Daniel Marín":
Hi, I think it would be very useful if users could specify a row in
the
worksheet to be assigned as variable labels and specify other row for
variable names. This would be appreciated if possible also in other
formats as csv. Ut would save a lot of time and problems.
For example enabling an option in the "labels" command.
I also would let the names and labels be of a longer number of
characteres, very especially the labels, as in many cases the
descriptions of series are considerably long.
Thank you for your attention, I ignore if this would be very complicated
to do, so forgive me if i ask for too much.
Best regards,
Hi, maybe it's just me, but I don't really understand the request. Could
you give an example of what you have in mind when you talk about
veriable labels vs. veraiable names?
thanks
sven
12:08 p.m.
On Sat, 25 May 2019, Sven Schreiber wrote:
Am 25.05.2019 um 17:46 schrieb "Daniel Marín":
> Hi, I think it would be very useful if users could specify a row in the
> worksheet to be assigned as variable labels and specify other row for
> variable names. This would be appreciated if possible also in other
> formats as csv. Ut would save a lot of time and problems.
> For example enabling an option in the "labels" command.
> I also would let the names and labels be of a longer number of
> characteres, very especially the labels, as in many cases the
> descriptions of series are considerably long.
> Thank you for your attention, I ignore if this would be very complicated
> to do, so forgive me if i ask for too much.
> Best regards,
Hi, maybe it's just me, but I don't really understand the request. Could
you give an example of what you have in mind when you talk about
veriable labels vs. veraiable names?
I think that what Daniel has in mind is a csv file like this:
<csv>
x,y
"first label","second label",
2,3
4,5
0,1
</csv>
At present, reading a file like that is possible, but involves a bit of
scripting (I can provide an example if needed).
-------------------------------------------------------
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
-------------------------------------------------------
12:30 p.m.
Am 25.05.2019 um 18:14 schrieb "Daniel Marín":
What I want is to import data from Excel files, LibreOffice files,
and
be able to specify the row in the Excel worksheet where the names of the
variables are, and the row where the descriptive labels are,
the same way you can specify in what sheet are the data to import and in
what row and column you want to begin importing.
OK, I see. I think the request makes sense, basically merging the
"labels --from-file=..." command into the open command.
However, it seems to me not many such files exist in the wild. Typically
I think the descriptions are in a header which is either un-formatted or
is actually arranged vertically. Can you show an example where you get
data with the descriptions in a row, aligned with variable names?
And let both things (names and descriptive labels) be of a longer
number
of characteres as usually they are really longer.
Right now I'm not sure what the limit for labels is. Apparently you have
run into space constraints?
(With respect to the names, we have recently discussed the issue of the
max character length on this list.)
Also it would be great if this could be done also for other formats
of
files to import, as .csv or .R.
What was the original format you had in mind, .xls(x) ?
With respect to .R, I remember some discussion that those files can
follow different specifications and hence are not directly supported,
but I don't remember the details right now.
So, as to avoid the pain of having to construct a .txt file with the
descriptive labels,
As I said, I think your request makes sense. However, are you aware of
the transpose function in Excel (and other spreadsheet programs)? You
just would have to:
1. copy the row with the labels (descriptions)
2. paste it (in a new sheet, for example) in a transposed arrangement
(column)
3. then copy&paste it into a plain text file
So this doesn't look like a lot of pain to me. Of course, if you have
many such files, things may look different.
thanks
sven
2:53 a.m.
On Sat, 25 May 2019, Sven Schreiber wrote:
Am 25.05.2019 um 18:14 schrieb "Daniel Marín":
> What I want is to import data from Excel files, LibreOffice files, and
> be able to specify the row in the Excel worksheet where the names of the
> variables are, and the row where the descriptive labels are [...]
Right now I'm not sure what the limit for labels is. Apparently
you have
run into space constraints?
In current gretl there's no limit on the length of the descriptive
label for a series. We made this change a couple of months (?) ago, in
the first instance to accommodate long descriptions coming from
dbnomics.
(There is of course a limit of 31 bytes on names of variables, and
that's not going to change.)
Daniel might want to consider batch conversion of Excel or ODS files
to CSV using soffice or gnumeric. Then gretl's text processing
facilities could be brought to bear on data files that contain
descriptive labels as well as series names.
Allin
4:55 a.m.
On Sun, 26 May 2019, Allin Cottrell wrote:
Daniel might want to consider batch conversion of Excel or ODS
files to CSV using soffice or gnumeric. Then gretl's text
processing facilities could be brought to bear on data files
that contain descriptive labels as well as series names.
Yes. I've found the "ssconvert" utility from gnumeric very fast and
reliable.
Having said this, it would be nice to have a few more options to the open
command, such as the possibility of skipping some initial lines, or
designating a certain line as the one containing labels, and so on.
-------------------------------------------------------
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
-------------------------------------------------------
12:58 p.m.
Am 26.05.2019 um 17:20 schrieb "Daniel Marín":
I've tried to install the new 2019 version I downloaded of Gretl
from
links on your web page, but I'm in UBUNTU and instead, it reinstsalled
the 2016 version I had... is there a way to install it in Ubuntu, maybe
it has rejected this new version becouse some issue with dependences...
Yes, I
agree it is a common problem with Linux distros that it's
sometimes difficult to upgrade only a single package. You have various
options:
1) I believe the latest Ubuntu (19.04) ships with gretl 2019a. But I
understand you may not want to upgrade your system to a one-month-old
distro release.
2) Compile from source: This is relatively easy under Debian-based
distros such as Ubuntu. See appendix C in the user guide.
3) [not something you can do] If somebody had a PPA-style repository to
provide more recent gretl versions for Ubuntu/Mint/Debian, that would be
cool. But it involves some work, of course.
cheers
sven
4:44 p.m.
Am 26.05.2019 um 22:11 schrieb "Daniel Marín":
Hi, Sven, thanks for your advice about compiling the program, I
don't
know if doing it, this will provoke any conflict with preexistent
libraries and dependences... is it safe?
Yes, it is "reasonably" safe,
because all the required software for
compilation is pulled in via the native package management system, the
only exception being gretl itself.
(The download of the requirements can be large.)
However, uninstalling the self-compiled version later is more
complicated than with the pre-compiled package. The hope is that you
don't need to do that...
cheers
sven
2002
days inactive
2003
days old
14 comments
4 participants
participants (4)
-
"Daniel Marín" -
Allin Cottrell -
Riccardo (Jack) Lucchetti -
Sven Schreiber