Hi Tom,
I am an avid R user and forgot about the spgrass6 addon. I have linked to grass
through R and read in my rasters with the following script. The process worked
beautifully and I appreciate you reminding me of the package and process!!!
Script for reading the rasters
#######
require(spgrass6)
require(EMCluster)
require(rgdal)
initGRASS("C:/Program Files (x86)/GRASS GIS 7.0.svn",
gisDbase="C:/Users/brown782/Desktop",location="grass_data",mapset ="PERMANENT",
override=T)
require(spgrass6)
require(EMCluster)
require(rgdal)
rasters <- readRAST6(c("slope_pct", "wetness_indx", "nomralized_height"),
cat=c(T, F, F), useGDAL=T)
#####
I decided to use the EMCluster package in R instead of the cmeans package.
Yesterday, I only tried to work in the R interface with the raster and
cmeans("e1071") packages and read in the grids as ascii files. My issue was
that when I stacked the grids and tried using cmeans it seemed that the fuzzy
cluster process did not seem to understand that my stacked grids were spatial
data.
So now that I am using the EMCluster package R is giving me an error that the
"rasters" object needs to be a matrix. I used the following script to convert
the data to a matrix.
rasters2 = as.matrix(rasters)
It seemed to work but R gave me the below warning that the Spatial.Data.Frame
######
In as.matrix.SpatialGridDataFrame(rasters) :
as.matrix.SpatialPixelsDataFrame uses first column;
pass subset or [] for other columns
#####
It seems that the as.matrix function somehow combined the three rasters into
one because when I looked at the dimensions of the newly formed matrix it only
gave me one set of dimensions. All of the rasters have the same dimensions.
Where the Dimensions are below for the three rasters and the newly formed
matrix.
slope_pct = raster:
rows: 6750
columns 9152
wetness_indx = raster:
rows: 6750
columns 9152
normalized_height = raster
rows: 6750
columns 9152
rasters2 = Matrix
rows: 6750
columns 9152
can I convert the three individual grids to their own matrix then in run the
EMCluster with the following script?
results = simple.init(slope_pct, wetness_indx, normalized_height, nclass = 5)
results = simple.init(slope_pct + wetness_indx + normalized_height, nclass = 5)
I tried the second above option and it failed to produce any kind of meaningful
cluster it just gave me a short list of NA and NAN Errors.
Below I used the following code for the EMCluster package with the "rasters2"
matrix but would like to replace it with either of the codes above?
results = simple.init(rasters2, nclass = 5)
This may be something I need to post in the R discussion forum.
Thanks for all of your help!
Robert Brown
----- Original Message -----
From: "Thomas Adams" <[email protected]>
To: "Robert A Brown" <[email protected]>
Cc: [email protected], "Markus Metz" <[email protected]>
Sent: Thursday, April 11, 2013 1:30:25 PM
Subject: Re: [GRASS-user] Fuzzy K-mean
Robert
I would recommend a combination of GRASS and R ( http://cran.r-project.org/ ).
R has fuzzy K-mean cluster capability, using the R addon package spgrass6 to
read/write to/from GRASS and R. Also, see this:
http://www.r-bloggers.com/fuzzy-clustering-with-fanny/ and this:
http://stats.stackexchange.com/questions/31906/fuzzy-k-means-cluster-sizes as
starting points.
Tom
On Thu, Apr 11, 2013 at 12:48 PM, Robert A Brown < [email protected] > wrote:
Hello All,
I am trying to find a way to process a fuzzy K-mean cluster using multiple
rasters/grids in GRASS. Is anyone aware of a grass add-on or another simple
process I can use to accomplish this?
Thanks,
Robert Brown
PhD Student
Purdue University
Department of Agronomy
_______________________________________________
grass-user mailing list
[email protected]
http://lists.osgeo.org/mailman/listinfo/grass-user
_______________________________________________
grass-user mailing list
[email protected]
http://lists.osgeo.org/mailman/listinfo/grass-user
