Hi all, I am sorry if this is a very basic quesion, but I have no experience
with analyzing spatial data and could not find the right function/package
quickly. Any hints would be much appreciated. I have a matrix of spatial
point patterns like the one below and want to find the number of independent
components (if that's the right term) in that matrix (or in that image).
x=matrix(c(0,1,0,0,0,
0,1,1,0,0,
0,0,0,0,0,
0,0,0,1,0,
0,0,0,1,0),nrow=5)
image(x)
I can find the number of populated points easily
table(x) #or more generally
sum(x!=0)
But I want to find the number of independent components. The answer in this
example should be 2. There are three criteria to the function I am seeking:
1. Points that have a neighboring nonzero point should be counted as one
contiguous component.
2. The function should respect that the matrix is projected on a torso. That
is, points in the leftmost column border points in the rightmost column and
points in the top row border points in the bottom row (if they are
contiguous when you wrap the image around a cylinder).
3. The function should be fast/efficient since I need to run this over
thousands of images/matrices.
Is anyone aware of an implementation of such a function?
Thanks much for your help,
Daniel
--
View this message in context:
http://r.789695.n4.nabble.com/Spatial-number-of-independent-components-tp2262018p2262018.html
Sent from the R help mailing list archive at Nabble.com.
I am sure this can be written in a much more elegant and faster code. One way I can think of, is to modify cell2nb code and create a new function that creates the neighbour lists of only cells that are not 0. These are best directed to R-sig-Geo list. However, the following might work. library(spdep) library(igraph)> x=matrix(c(0,1,0,0,0, > 0,1,1,0,0, > 0,0,0,0,0, > 0,0,0,1,0, > 0,0,0,1,0),nrow=5)a <- nb2mat(cell2nb(nrow(x),ncol(x)), style="B", torus="TRUE") g <- delete.vertices(graph.adjacency(a), which(x!=1)-1) plot(g) clusters(g) Nikhil --- Nikhil Kaza Asst. Professor, City and Regional Planning University of North Carolina nikhil.list at gmail.com On Jun 20, 2010, at 7:17 PM, Daniel Malter wrote:> > Hi all, I am sorry if this is a very basic quesion, but I have no > experience > with analyzing spatial data and could not find the right function/ > package > quickly. Any hints would be much appreciated. I have a matrix of > spatial > point patterns like the one below and want to find the number of > independent > components (if that's the right term) in that matrix (or in that > image). > > x=matrix(c(0,1,0,0,0, > 0,1,1,0,0, > 0,0,0,0,0, > 0,0,0,1,0, > 0,0,0,1,0),nrow=5) > > image(x) > > I can find the number of populated points easily > > table(x) #or more generally > sum(x!=0) > > But I want to find the number of independent components. The answer > in this > example should be 2. There are three criteria to the function I am > seeking: > > 1. Points that have a neighboring nonzero point should be counted as > one > contiguous component. > > 2. The function should respect that the matrix is projected on a > torso. That > is, points in the leftmost column border points in the rightmost > column and > points in the top row border points in the bottom row (if they are > contiguous when you wrap the image around a cylinder). > > 3. The function should be fast/efficient since I need to run this over > thousands of images/matrices. > > Is anyone aware of an implementation of such a function? > > Thanks much for your help, > Daniel > -- > View this message in context: http://r.789695.n4.nabble.com/Spatial-number-of-independent-components-tp2262018p2262018.html > Sent from the R help mailing list archive at Nabble.com. > > ______________________________________________ > R-help at r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code.
Hi, thanks much. This works in principle. The corrected code is below: a <- nb2mat(cell2nb(nrow(x),ncol(x),torus=T), style="B") g <- delete.vertices(graph.adjacency(a), which(x!=1)-1) plot(g) clusters(g) the $no argument of the clusters(g) function is the sought after number. However, the function is very slow, and my machine runs out of memory (1G) for a 101 by 101 matrix. Daniel -- View this message in context: http://r.789695.n4.nabble.com/Spatial-number-of-independent-components-tp2262018p2262090.html Sent from the R help mailing list archive at Nabble.com.
I just updated spdep and I see that as.spam.listw works. Below is
sessionInfo
Furthermore, it may be straightforward to condense the adjacency
matrix *before* converting to graph which may help a little bit. You
can profile the code and see which part needs speeding up.
library(spdep)
library(igraph)
x=matrix(c(0,1,0,0,0,
0,1,1,0,0,
0,0,0,0,0,
0,0,0,1,0,
0,0,0,1,0),nrow=5)
a <- as.spam.listw(nb2listw(cell2nb(nrow(x),ncol(x),torus=T),
style="B"))
ind <- which(x>0)
b <- a[ind, ind]
g1 <- graph.adjacency(b)
clusters(g1)$no
---
R version 2.11.1 (2010-05-31)
i386-apple-darwin9.8.0
locale:
[1] en_US.UTF-8/en_US.UTF-8/C/C/en_US.UTF-8/en_US.UTF-8
attached base packages:
[1] stats graphics grDevices utils
[5] datasets methods base
other attached packages:
[1] igraph_0.5.3 spam_0.22-0
[3] spdep_0.5-11 coda_0.13-5
[5] deldir_0.0-12 maptools_0.7-34
[7] foreign_0.8-40 nlme_3.1-96
[9] MASS_7.3-6 Matrix_0.999375-39
[11] lattice_0.18-8 boot_1.2-42
[13] sp_0.9-64
loaded via a namespace (and not attached):
[1] grid_2.11.1 tools_2.11.1
Nikhil Kaza
Asst. Professor,
City and Regional Planning
University of North Carolina
nikhil.list at gmail.com