The centroids don't "represent" the clusters, though, and you can
construct arbitrary complex
clusterings where all the centroids are identical.
On 10/20/2017 01:08 PM, Sebastian Raschka wrote:
Independent from the implementation, and unless you use the 'centroid' or
'average linkage' method, cluster centroids don't need to be computed when
performing the agglomerative hierarchical clustering . But you can always
compute it manually by simply averaging all samples from a cluster (for each
feature).
Best.
Sebastian
On Oct 20, 2017, at 9:13 AM, Sema Atasever <s.atase...@gmail.com> wrote:
Dear scikit-learn members,
I am using SciPy's hierarchical agglomerative clustering methods to cluster a
1000 x 22 matrix of features, after clustering my data set with
scipy.cluster.hierarchy.linkage and and assigning each sample to a cluster,
I can't seem to figure out how to get the centroid from the resulting clusters.
I would like to extract one element or a few out of each cluster, which is the
closest to that cluster's centroid.
Below follows my code:
D=np.loadtxt(open("C:\dataset.txt", "rb"), delimiter=";")
Y = hierarchy.linkage(D, 'ward')
assignments = hierarchy.fcluster(Y, 5, criterion="maxclust")
I am taking my matrix of features, computing the euclidean distance between
them, and then passing them onto the hierarchical clustering method. From
there, I am creating flat clusters, with a maximum of 5 clusters
Now, based on the flat clusters assignments, how do I get the 1 x 22 centroid
that represents each flat cluster?
Best.
<SciPy_python_codes.py><dataset.txt><assignments.out>_______________________________________________
scikit-learn mailing list
scikit-learn@python.org
https://mail.python.org/mailman/listinfo/scikit-learn
_______________________________________________
scikit-learn mailing list
scikit-learn@python.org
https://mail.python.org/mailman/listinfo/scikit-learn
_______________________________________________
scikit-learn mailing list
scikit-learn@python.org
https://mail.python.org/mailman/listinfo/scikit-learn
