Clustering with Scipy in Python? (hierarchical clustering)

Question

I'm a bit confused about the clusering with Scipy in Python. Here is my sourcecode:

import scipy.spatial.distance as dist
import numpy, scipy

dataMatrix = numpy.array(matrix)
distMatrix = dist.pdist(dataMatrix, 'euclidean')
distSquareMatrix = dist.squareform(distMatrix)

Y = scipy.cluster.hierarchy.linkage(distSquareMatrix, method='complete')

Do I have to use the 'distMatrix' or the squareform 'distSquareMatrix' as input for the clustering? Because I saw both methods in other posts. But the output is different. Now I'm not sure what I have to choose.

andrewsi · Accepted Answer · 2013-11-09 18:01:04Z

4

You need to pass the distance matrix in condensed form without transforming it with squareform. The squareform function is useful if you want to manipulate the distance matrix yourself more easily as a 2D array. The scipy.cluster.hierarchy functions use the condensed form for the purpose of saving roughly a factor of two in memory.

I hope this helps.

edited Nov 9, 2013 at 18:01

andrewsi

10.7k132 gold badges37 silver badges52 bronze badges

answered Nov 9, 2013 at 17:40

user2973497

412 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

Joko Over a year ago

Thanks for your answer! That means I have to use 'distMatrix' in the linkage function...

Collectives™ on Stack Overflow

Clustering with Scipy in Python? (hierarchical clustering)

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related