Thank you very much, John and Alan for your quick response and feedback. Your comments are both clarifying and encouraging. I'll meet with a couple of professors in a local university to see if they are willing to give it a try this next January. Any additional information, as well as Jose Marchesi response will be welcome.
Jose I Dominguez
On 4 dic. 2018 04:51, John Darrington
wrote:
A task like adding a SAVE subcommand to QC is something that I think a
competent compter scientce student could complete within one semester. But
Alan is right, it would require that student to be well versed in C. And it
would also need an understanding of some PSPP internals - this undertstanding
could perhaps be imparted by way of mentoring.
Possibly, a task like this might be suitable for the Google summer of code. I
know that GNU normally participates in that, so it could be done under the GNU
auspecies. I'll ask Jose Marchesi who normally organises it what the procedure
would be.
So far as building the GUI under old versions of GNU/Linux is concerned, I
don't have any better suggestions than what Alan has mentioned, except perhaps
to move either to a newer OS or one such as GuixSD which allows multiple versions
of libraries to be installed concurrently.
J'
On Mon, Dec 03, 2018 at 09:01:44PM -0600, Alan Mead wrote:
Jose,
Maybe this would be better on the pspp-dev list, but it comes up every
so often. I added the support (with a lot of hand-holding by John
Darrington) for /PRINT to the Quick Cluster and I intended to add /SAVE
as well. Adding /PRINT was not very hard, but adding /SAVE will require
that your students not only know C but also understand how PSPP manages
data (which makes sense, you need to add a new variable to the dataset).
I was going to model the QC /SAVE off the /SAVE feature implemented for
regression. I'm sure that's possible, but it was less straightforward
than I'd hoped and required more effort than I had at the time.
Another issue for me was that the full gui pspp (i.e., PSPPIRE) cannot
be compiled on the Linuxes I typically use (CentOS 6 and 7). It looks
like Fedora 28 could do it if you can install the GNU
spread-sheet-widget (it's not clear, because configure said it needed
Cairo 1.5 but F29/F29 only have 1.15, but the errors ceased as I
installed more dependencies); John Darrington wrote the spreadsheet
widget, so maybe he could help you. I haven't been following the thread,
but that package is available in Debian unstable. I think most of the
PSPP developers use Debian unstable.
Of course, you can always compile it without a gui and run it on the
command line. That works great for me, but I guess most pspp users only
use the gui.
So, I don't want to discourage you (quite the opposite), but your
students would need to be comfortable with these issues. If they were
comfortable with these issues, I'm sure they could add this feature as a
class project in a semester. It's a fairly small amount of code.
One other thing: I had to go through an assignment/disclaimer process to
assign my copyrights; I guess your students would have to do so as well.
They would, obviously, have to agree to this and they should start that
process near the beginning of the semester because it took about a month
for my form to be processed and executed.
-Alan
On 12/3/2018 4:53 PM, Jose Ignacio Dominguez wrote:
>
> I find PSPP very useful, and I'm sure it can be a perfect substitute
> for SPSS (and many other packages) in its current procedures.
>
> I do have one suggestions which I'd think shouldn't be that hard to
> solve. While there is a command to Save Zvalues in the Descriptives
> procedure, there is not a Save command which would be extremely useful
> in other procedures. Factor (save factor scores), Quick Cluster (save
> cluster membership) Logistic Regression (save predicted, and save
> probability). Particularly, the save cluster membership is mandatory,
> given that we run the procedure to assign elements to groups. I fin
> there is a Print command that does a similar task, so saving a new
> variable should not be that hard.
>
> I have other suggestions that I presume involve a much harder task,
> regarding additional procedures available in current SPSS versions:
> Hierarchical Cluster, Discriminant Analysis, Correspondence Analysis,
> and Decision Trees.
>
> ??Are this tasks too hard to cover, or do you think a small team of
> students could do the job as a class project in the course of a semester?
>
> --
>
> _______________________________________________
> Pspp-users mailing list
> address@hidden
> https://lists.gnu.org/mailman/listinfo/pspp-users
--
Alan D. Mead, Ph.D.
President, Talent Algorithms Inc.
science + technology = better workers
http://www.alanmead.org
"You're an interesting species. An interesting mix.
You're capable of such beautiful dreams, and such
horrible nightmares. You feel so lost, so cut off,
so alone, only you're not. See, in all our
searching, the only thing we've found that makes
the emptiness bearable, is each other."
-- Carl Sagan, Contact
_______________________________________________
Pspp-users mailing list
address@hidden
https://lists.gnu.org/mailman/listinfo/pspp-users
--
