[Sugar-devel] The quest for data

Sameer Verma sverma at sfsu.edu
Fri Jan 3 14:16:37 EST 2014


On Fri, Jan 3, 2014 at 4:15 AM, Martin Abente
<martin.abente.lahaye at gmail.com> wrote:
> Hello Sameer,
>
> I totally agree we should join efforts for a visualization solution, but,
> personally, my main concern is still a  basic one: what are the important
> questions we should be asking? And how can we answer these questions
> reliably? Even though most of us have experience in deployments and their
> needs, we are engineers, not educators, nor decision makers.
>

Agreed. It would be helpful to have a conversation on what the various
constituencies need (different from want) to see at their level. The
child, the parents/guardians, the teacher, the
principal/administrator, and educational bureaucracy. We should also
consider the needs of those of us who have to fundraise by showing
progress of ongoing effort.

> I am sure that most of our collection approaches cover pretty much the
> trivial stuff like: what are they using, when are they using it, how often
> they use it, and all kind of things that derive directly from journal
> metadata. Plus the extra insight that comes when considering different
> demographics

True. Basic frequency counts such as frequency of use of activities,
usage by time of day, day of week, scope of collaboration are a few
simple one. Comparison of one metric vs the other will need more
thinking. That's where we should talk to the constituents.

>
> But, If we could also work together on that (including the trivial
> questions), it will be a good step forward. Once we identify these questions
> and figure out how to answer them, it would be a lot easier to think about
> visualization techniques, etc.

If the visualization subsystem (underlying tech pieces) are common and
flexible, then we can start with a few basic templates, and make it
extensible, so we can all aggregate, collate, and correlate as needed.
I'll use an example that I'm familiar with. We looked at CouchDB for
two reasons: 1) It allows for sync over intermittent/on-off
connections to the Internet and 2) CouchDB has a "views" feature which
provides selective subsets of the data, and the "reduce" feature does
aggregates. The actual visual is done in Javascript. Here's the
example Leotis had at the OLPC SF summit
(http://108.171.173.65:8000/).
>
> What you guys think?
>

A great start for a great year ahead!

> Saludos,

cheers,
> tch.
Sameer


More information about the Sugar-devel mailing list