PSPP-BUG: Re: [bug #30008] Aggregate data

[John Darrington]

On Sat, Aug 07, 2010 at 08:43:06AM +0200, Fredrik Clementz wrote:

     Some feedback on the aggregate command! It seems most stuff is in place.
     Uncertain if you aim for a practical look or a cloned look or your own 
look.

A practical look is of primary concern.  We try to follow the Gnome 
"Human Interface Guildlines" http://library.gnome.org/devel/hig-book/nightly/
However since many of our users are familiar with the SPSS interface, sometimes
we depart from the HIG when we think it'll make those users too confused.
     
     
     1. The default size of the variable listbox (at least in windows) is way 
too
     small. It is so narrow that the first thing that has to be done to work 
with
     this window is expanding it in width. Therefor may I suggest that it should
     be wider as default so that one doesn't have to expand it every time?
     Possibly making the entire window just a little big bigger as well.

I pushed a fix setting the panel to 50% of its maximum size.
     
     2. THe most common function I use aggregate for is simply looking for
     "duplicates" or similar. Now how this is done is simply using a number of
     variables that I push in as break variables. THen I usually tick in "Number
     of cases" and run it. However, in PSPP, there is no "Number of cases box"
     and just adding the variables to break won't allow me to press ok.  I
     believe this is the most common use of this function which is likely why
     SPSS added this quick tickbox instead of fiddling around with the more
     daring "functions" that more experienced users deal with (Ehrm, not to
     mention more experienced tend to fall to Syntax mode). I would suggest
     relooking how this is done to not make it differ from SPSS to much as I
     really believe there approoach is better than ours at the moment.

What does the "number of cases" checkbox do?  What syntax does it produce?
     
     3. If we'd really like to go a bit beyond SPSS which I personally would
     enjoy, one of the functions I would appreciate is to add the function
     "random". So when aggregating a dataset and there are 2 similar cases, how
     about choosing a random value from these. Would be really useful.

I suppose it's possible.  If this is really important to you, please raise it
as a wishlist bug. Make sure you give a detailed description of how it would 
work.
(Eg, what constitutes a "similar" case).
     
     4. Assuming we'll keep the way functions are added now (I really see the
     benefit of efficiency not having popups over popups), may I suggest doing a
     similar grouping of options witha  header as done in SPSS. Like -Summary- 
or
     -Specific- Etc.... Food for thought, this one is discussable I agree.

That's certainly possible.  A treeview could replace the combobox and then all
the items could be collapsed or expanded as desired.
     
     5. I note that we have "replace the current dataset with the aggregated
     variables" while SPSS has "create a new dataset containing only the
     aggregated variables" with the option of giving it a name. Now I would
     believe that these 2 are similar? Just that our current name implies
     removing the current dataset in favour of the aggregated. Think some
     clarification would be nice.


Currently, PSPP doesn't support multiple datasets.  So the current wording
is correct.  When (if?) we implement multiple datasets, then a small adjustment
to the interface will be appropriate.
     
J'

-- 
PGP Public key ID: 1024D/2DE827B3 
fingerprint = 8797 A26D 0854 2EAB 0285  A290 8A67 719C 2DE8 27B3
See http://pgp.mit.edu or any PGP keyserver for public key.

signature.asc
Description: Digital signature