[R-sig-Geo] Factor Analysis using R and grass

[Agustin Lobo]

Brian,

PCA does not care about raster or vector. All you need is
a table of individuals x variables. Whether you get that
from a vector or a raster does not matter at all. Instead,
you must be careful with what your data mean. In particular,
you mention centroids. If those are centroids of polygons
which do not have the same area and your centroid
is the mean of a surface-dependent variable
, you must be aware of the fact
the PCA of A and B below is not the same (B has the first row of A 3 
times, the second once
and the third 5, i.e the area of the 3 polygons would be 3, 1, 5 units)

 > A
[,1] [,2] [,3] [,4]
[1,] 0.7142542 2.4097913 0.36845987 -0.08292664
[2,] -0.1718578 -1.2655390 -0.01597638 -0.51156564
[3,] 0.3774107 -0.4042545 0.35403702 -1.01637522
 > B
[,1] [,2] [,3] [,4]
[1,] 0.7142542 2.409791 0.36845987 -0.08292664
[2,] 0.7142542 2.409791 0.36845987 -0.08292664
[3,] 0.7142542 2.409791 0.36845987 -0.08292664
[4,] -0.1718578 -1.265539 -0.01597638 -0.51156564
[5,] 0.7142542 2.409791 0.36845987 -0.08292664
[6,] 0.7142542 2.409791 0.36845987 -0.08292664
[7,] 0.7142542 2.409791 0.36845987 -0.08292664
[8,] 0.7142542 2.409791 0.36845987 -0.08292664
[9,] 0.7142542 2.409791 0.36845987 -0.08292664

Also, note that PCA is not spatial at all.

Agus

Brian Cooper wrote:
> Thanks for the response; the data sets are centroid based with 20 to 30
> variables per centroid. What has been suggested will work with rasters but
> not with vectors. What I need to know it is possible to conduct a PCA on
> vector data sets and store the results as additional variables. I am
> interested in the impact of the nearest neighbour on the particular score. I
> work with human services planning data and am looking at developing more
> effective measures for local area planning. Theoretically speaking a
> geo-statistical approach will give a truer result that current aspatial
> approaches.
>
> Brian Cooper
>
> -----Original Message-----
> From: r-sig-geo-bounces at stat.math.ethz.ch
> [mailto:r-sig-geo-bounces at stat.math.ethz.ch] On Behalf Of Brian Cooper
> Sent: Thursday, 26 March 2009 12:27 AM
> To: r-sig-geo at stat.math.ethz.ch
> Subject: [R-sig-Geo] Factor Analysis using R and grass
>
> I am new to both R and Grass. I need to duplicate the Principal Component
> Analysis approach used in SPSS with GRASS and R. Is this possible?
>  
> brian Cooper
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
> _______________________________________________
> R-sig-Geo mailing list
> R-sig-Geo at stat.math.ethz.ch
> https://stat.ethz.ch/mailman/listinfo/r-sig-geo
>
>