Python k-means algorithm

user_122032 0
Viewed 89804

I am looking for Python implementation of k-means algorithm with examples to cluster and cache my database of coordinates.

k-means cluster-analysis algorithm python
8 Answers

SciPy's kmeans2() has some numerical problems: others have reported error messages such as "Matrix is not positive definite - Cholesky decomposition cannot be computed" in version 0.6.0, and I just encountered the same in version 0.7.1.

For now, I would recommend using PyCluster instead. Example usage:

>>> import numpy
>>> import Pycluster
>>> points = numpy.vstack([numpy.random.multivariate_normal(mean, 
                                                            0.03 * numpy.diag([1,1]),
                                                            20) 
                           for mean in [(1, 1), (2, 4), (3, 2)]])
>>> labels, error, nfound = Pycluster.kcluster(points, 3)
>>> labels  # Cluster number for each point
array([1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 0, 0, 0,
       0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 2, 2, 2, 2, 2,
       2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2], dtype=int32)
>>> error   # The within-cluster sum of distances for the solution
1.7721661785401261
>>> nfound  # Number of times this solution was found
1
user_174980
Related