pyclustering
0.10.1
pyclustring is a Python, C++ data mining library.

Enumeration of initialization types for ExpectationMaximization algorithm. More...
Static Public Attributes  
int  RANDOM_INITIALIZATION = 0 
Means are randomly taken from input dataset and variance or covariance is calculated based on spherical data that belongs to the chosen means.  
int  KMEANS_INITIALIZATION = 1 
Two step initialization. More...  
Enumeration of initialization types for ExpectationMaximization algorithm.

static 
Two step initialization.
The first is calculation of initial centers using KMeans++ method. The second is KMeans clustering using obtained centers in the first step. Obtained clusters and its centers are used for calculation of variance (covariance in case of multidimensional) data.