[R-sig-Geo] spatial clustering - land usage - skater?

[Georg Ruß]

On 09/03/11 10:40:00, Mathieu Rajerison wrote:
> I've got a dataset of points that covers a city

Okay, is that a spatial points data frame?

> Each point takes 4 values corresponding to area of land usage. The first one
> is for ground, the second is built, the third is wood, the fourth water.

Here it looks like each point in the spatialpoints data frame has a vector
with four entries. Are those binary entries? I.e., if it's a point for
'ground', is the vector c(1,0,0,0) since the other entries must be zero?
Or is the vector just having one entry, coding one of the four values for
"ground", "built", "wood", "water"?

> I'd like to seggregate components corresponding to each of land usage:
> ground / built / wood / water.

I guess once we get the problem and the data structure sorted out, it
shouldn't be one hell of a problem.

> There may be several components for each category. For example, 2 for
> ground, 3 for water and this might be ajusted by the user.

Hmm, here you're talking about categories, which you didn't mention
before. Please clarify, maybe give us a hint on the data structure.

> What is the most appropriate function to accomplish that?
> I considered using spdep package with skater? Is it the most appropriate?

I wouldn't comment on that unless we've had the problem clarified. So
please help us. If you have, post reproducible code fragments.

Regards,
Georg.
--
Research Assistant
Otto-von-Guericke-Universität Magdeburg
research at georgruss.de
http://research.georgruss.de