pyclustering.cluster.cure.cure.get_clusters() function Problem

[PrachiPrakash]

Actually get_clusters() function for cure algorithm return a list of clusters but each cluster is not a list of index but a list of points which is not consistent with other implementations.

Thanks

[annoviko]

Hello, @PrachiPrakash,

This is known issue, but anyway thank you for the reporting, I agree with you that get_clusters() should be unified for all clustering algorithms in the library and cure::get_clusters() should return indexes instead of points. I will update the method for Python and for C/C++ implementations.

[annoviko]

@PrachiPrakash, changes are delivered to branch 0.7.dev, they will be available in the next 0.7.3 release.

[himanshu94]

The same problem exists for X means.

Thanks

[annoviko]

Hello, @himanshu94,

Could you please explain or share the code where you have found this issue using X-Means?

There is tests for Python and C/C++ implementation and both returns indexes instead of points. I have written an example that demonstrates it:

from pyclustering.cluster.xmeans import xmeans;
from pyclustering.cluster.center_initializer import kmeans_plusplus_initializer;

from pyclustering.utils import read_sample;

from pyclustering.samples.definitions import SIMPLE_SAMPLES;

# Read dataset 'SAMPLE_SIMPLE2'
sample = read_sample(SIMPLE_SAMPLES.SAMPLE_SIMPLE2);
initial_centers = kmeans_plusplus_initializer(sample, 3).initialize();

# Use Python implementation
xmeans_instance = xmeans(sample, initial_centers);
xmeans_instance.process();
clusters = xmeans_instance.get_clusters();

# Display allocated clusters
print(clusters);

# Use C/C++ implementation
xmeans_instance = xmeans(sample, initial_centers, ccore=True);
xmeans_instance.process();
clusters = xmeans_instance.get_clusters();

# Display allocated clusters
print(clusters);

Output of the code is

[[0, 1, 2, 3, 4, 5, 6, 7, 8, 9], [15, 16, 17, 18, 19, 20, 21, 22], [10, 11, 12, 13, 14]]
[[0, 1, 2, 3, 4, 5, 6, 7, 8, 9], [15, 16, 17, 18, 19, 20, 21, 22], [10, 11, 12, 13, 14]]

There option how to identify what kind of representation is used (it returns type_encoding.CLUSTER_INDEX_LIST_SEPARATION):

representaion = xmeans_instance.get_cluster_encoding();

https://codedocs.xyz/annoviko/pyclustering/classpyclustering_1_1cluster_1_1encoder_1_1type__encoding.html

[himanshu94]

Thanks @annoviko . The link you provided was helpful.