[OSGeodata] Tie-in to Education Committee

[Tyler Mitchell]

On Thursday 13 April 2006 08:56, Ned Horning wrote:
> What kind of global dataset are you wishing for? I think I'm missing the
> point because when I want a dataset for a tutorial or something I think
> about the point(s) I want to get across and then I go get some data. There
> are lots of great data sources out there and many are free. In my mind
> generating datasets is relatively easy (as long as the data I want is
> accessible) so clearly I'm missing something. For me the hard part is
> figuring out what I want to illustrate.

Hi Ned,
I know exactly what you are saying but I think you are exceptional :)  The gap 
that I would like to see filled is for those who are either ignorant of good 
datasets or unable/unwilling to spend time compiling them - yet who still 
want to teach OSGeo material.  For most of us who want to demonstrate 
functionality of a product, there is little benefit in spending time to pull 
together customised tutorial datasets.

> I wouldn't be surprised if we put together some dataset descriptions (I
> can't think of a single "ideal" data set) that we would find many already
> exist and can be freely used. Those that don't exist can be assembled
> easily. Would it make sense to create a wish list of assorted tutorial
> datasets? If I could better understand what people want I think I'd better
> understand the problem and might have some ideas.

Yes, I think it would make sense to create a wish list of tutorial datasets, 
categorised as to content or geographic extent, application, etc.  But I see 
addressing that needing longer term timelines to address.

I'm not trying to find a single ideal data set, but a few that are simply good 
enough for demonstrating software capabilities.  For example, a lo-res global 
raster and country boundaries would be fine for showing off the features of a 
WMS and a WFS.  Add in some more political boundaries or transportation data 
and we'd have a very simple, yet useful, package.  It could then also serve 
well as an input dataset for desktop applications to show how to do 
colour-theming, simple analysis, etc.

> For global data a canned dataset is sometimes less than ideal because
> people (especially kids) will want to see detail in their own community.
> This can be accomplished but not easily with a canned dataset.

If we had global coverage of community-level datasets, I wouldn't be aiming 
for something 'good enough' - I'd be aiming for getting my hands on that 
precious package ;)  Seriously though, it's always helpful to have a starting 
point, even if it isn't ideal.

If there is data available more locally and the tutor/teacher is able to 
assemble it, that makes sense.  I'm thinking of those situations where this 
is not the case.  And also so we can have some 'baseline' datasets that can 
be easily distributed as a common tutorial platform.  For example, the annual 
workshops that many of us lead at conferences, etc. don't usually have a 
focus on a certain geographic area, instead we show how to do things with the 
software.  Those are the kinds of scenarios I am thinking of.

Does that help fill in the picture for you?  Let me know if I'm talking in 
circles!

Talk to you later,
Tyler