R help - May 2014 - Using centers of hierarchical clustering for k-means

Hi,

i have the following problem: I am using k-means algorithm for clustering.
But instead of using randomized centers, I would like to use centers created
by hierarchical clustering. So I want to apply "hclust" on my data set
(in
this case the iris data), getting a solution by "cutree", calculating
the
means/centers of the resulting clusters and use these centers as starting
points for k-means clusterng. But I have no idea how I calculate the centers
of the clusters and how to use them as starting points for the k-means
algorithm.

Hope you can help. Thanks in advance.

Mario



--
View this message in context:
http://r.789695.n4.nabble.com/Using-centers-of-hierarchical-clustering-for-k-means-tp4690704.html
Sent from the R help mailing list archive at Nabble.com.

This should get you started
> set.seed(42)
> x <- matrix(rnorm(200, 25, 5), 40, 5)
> x.clus <- hclust(dist(x))
> x.g4 <- cutree(x.clus, 4)
> x.cent <- aggregate(x, list(x.g4), mean)
> x.km <- kmeans(x, x.cent[,-1])
> xtabs(~x.g4+x.km$cluster)
    x.km$cluster
x.g4  1  2  3  4
   1 10  0  1  0
   2  0 12  2  0
   3  0  2 10  0
   4  0  0  0  3

-------------------------------------
David L Carlson
Department of Anthropology
Texas A&M University
College Station, TX 77840-4352


-----Original Message-----
From: r-help-bounces at r-project.org [mailto:r-help-bounces at r-project.org]
On Behalf Of marioger
Sent: Friday, May 16, 2014 7:29 AM
To: r-help at r-project.org
Subject: [R] Using centers of hierarchical clustering for k-means

Hi,

i have the following problem: I am using k-means algorithm for clustering.
But instead of using randomized centers, I would like to use centers created
by hierarchical clustering. So I want to apply "hclust" on my data set
(in
this case the iris data), getting a solution by "cutree", calculating
the
means/centers of the resulting clusters and use these centers as starting
points for k-means clusterng. But I have no idea how I calculate the centers
of the clusters and how to use them as starting points for the k-means
algorithm.

Hope you can help. Thanks in advance.

Mario



--
View this message in context:
http://r.789695.n4.nabble.com/Using-centers-of-hierarchical-clustering-for-k-means-tp4690704.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.

Well, you could use the group membership from the clustering along
with, for instance, aggregate() to get the mean values for each
cluster, and pass those to kmeans() using the centers argument as
described in the help file.

Unless you want medoids instead of centroids, since you didn't specify?

If you post a reproducible example including the code for the parts
you know how to do, you might get someone willing to fill in the code
for the unknown parts.

Sarah

On Fri, May 16, 2014 at 8:29 AM, marioger <mario_wiegand at gmx.de>
wrote:
> Hi,
>
> i have the following problem: I am using k-means algorithm for clustering.
> But instead of using randomized centers, I would like to use centers
created
> by hierarchical clustering. So I want to apply "hclust" on my
data set (in
> this case the iris data), getting a solution by "cutree",
calculating the
> means/centers of the resulting clusters and use these centers as starting
> points for k-means clusterng. But I have no idea how I calculate the
centers
> of the clusters and how to use them as starting points for the k-means
> algorithm.
>
> Hope you can help. Thanks in advance.
>
> Mario
>
>
-- 
Sarah Goslee
http://www.functionaldiversity.org