Re: Setting cluster seeds for K-Means analysis in SPSS

From: Borut (borut_rajer_at_hotmail.com)
Date: 07/30/04


Date: 30 Jul 2004 00:42:09 -0700

First do a K-means, 2 cluster, select Write final as (select Options
(?) in the main window of the procedure) ...
You get the file.
Do the Hierarchical - 2 clusters and get the centroid data with Means
procedure.
Open the previously saved file, replace the data in SPSS and save.
Now use the K-means again but instead of Write final select Read
initial (or something like that).

If you try to create a data file directly it doesn't seem to work.
There is however a quicker way if you use the syntax, but I'm afraid
I've forgotten it.

reneeh63@insightbb.com (Renee) wrote in message news:<8820eb75.0407271305.3f9d16bf@posting.google.com>...
> Hello -
> I've done a heirarchical cluster analysis and decided on a 2 cluster
> solution. Now I'd like to use the seeds of that analysis for a
> K-Means cluster. In what form are the seeds saved via the
> heirarchical analysis and how do I input them so k-means can use them?
>
> Thanks much everyone!



Relevant Pages

  • Re: Setting cluster seeds for K-Means analysis in SPSS
    ... First do a K-means, 2 cluster, select Write final as (select Options ... Open the previously saved file, replace the data in SPSS and save. ... > I've done a heirarchical cluster analysis and decided on a 2 cluster ...
    (sci.stat.edu)
  • Re: Clustering von kategoriellen Daten
    ... > Florian Laws wrote: ... wieviele Cluster man haben will. ... Ich hatte k-Means nur mal so als Beispiel genommen, ... Next by Date: ...
    (de.sci.informatik.ki)
  • Re: Clustering wrong results in SQL 2005
    ... The default Expectation-Maximization algorithm, also known as soft clustering, actually assigns each case to each cluster with some probability, while K-means assigns each case to one cluster with probability 1. ... Maybe the SUPPORT in NODE_DISTRIBUTION could be helpful - I noticed it was higher than in NODE_SUPPORT but I don't know how it was calculated. ...
    (microsoft.public.sqlserver.datamining)
  • RE: Clustering Algorithms
    ... The "value" that is clustered for discrete attributes is thier probability of being in that cluster. ... Both K-means and EM handle the attributes the same way, but the calculations are somewhat different as would be expected. ... Are there any differences between the two algorithms EM and K-Means ...
    (microsoft.public.sqlserver.datamining)
  • Setting cluster seeds for K-Means analysis in SPSS
    ... I've done a heirarchical cluster analysis and decided on a 2 cluster ... K-Means cluster. ... In what form are the seeds saved via the ...
    (sci.stat.math)