[GRASS-dev] i.cluster/i.maxlik question

I’m trying out GRASS’s unsupervised classification functions. As I understand it, it is a 2 step process.

  1. run i.cluster to get cluster signatures
  2. run i.maxlik using the cluster signatures to generate the spatial clusters

I’ve varied the parameters several times, and AFAICT, setting the initial number of clusters also determines the final number. In other unsupervised clustering routines I’ve used, this is not the case. Because the clustering iterates until the convergence value is reached, the final number of clusters can be different from the initial number. Can anyone offer insight on this for GRASS?

Thanks
Michael


C. Michael Barton
Director, Center for Social Dynamics & Complexity
Professor of Anthropology, School of Human Evolution & Social Change
Head, Graduate Faculty in Complex Adaptive Systems Science
Arizona State University

voice: 480-965-6262 (SHESC), 480-965-8130/727-9746 (CSDC)
fax: 480-965-7671 (SHESC), 480-727-0709 (CSDC)

www: http://www.public.asu.edu/~cmbarton, http://csdc.asu.edu

As a followup, it looks like changing the minimum cluster size may reduce the number of final clusters, but the final number seem not to exceed the initial number.

Michael


C. Michael Barton
Director, Center for Social Dynamics & Complexity
Professor of Anthropology, School of Human Evolution & Social Change
Head, Graduate Faculty in Complex Adaptive Systems Science
Arizona State University

voice: 480-965-6262 (SHESC), 480-965-8130/727-9746 (CSDC)
fax: 480-965-7671 (SHESC), 480-727-0709 (CSDC)

www: http://www.public.asu.edu/~cmbarton, http://csdc.asu.edu

On Nov 26, 2015, at 2:03 PM, Michael Barton <michael.barton@asu.edu> wrote:

I’m trying out GRASS’s unsupervised classification functions. As I understand it, it is a 2 step process.

  1. run i.cluster to get cluster signatures
  2. run i.maxlik using the cluster signatures to generate the spatial clusters

I’ve varied the parameters several times, and AFAICT, setting the initial number of clusters also determines the final number. In other unsupervised clustering routines I’ve used, this is not the case. Because the clustering iterates until the convergence value is reached, the final number of clusters can be different from the initial number. Can anyone offer insight on this for GRASS?

Thanks
Michael


C. Michael Barton
Director, Center for Social Dynamics & Complexity
Professor of Anthropology, School of Human Evolution & Social Change
Head, Graduate Faculty in Complex Adaptive Systems Science
Arizona State University

voice: 480-965-6262 (SHESC), 480-965-8130/727-9746 (CSDC)
fax: 480-965-7671 (SHESC), 480-727-0709 (CSDC)

www: http://www.public.asu.edu/~cmbarton, http://csdc.asu.edu