Module: algorithms.clustering.utils


nipy.algorithms.clustering.utils.kmeans(X, nbclusters=2, Labels=None, maxiter=300, delta=0.0001, verbose=0, ninit=1)

kmeans clustering algorithm

X: array of shape (n,p): n = number of items, p = dimension

data array

nbclusters (int), the number of desired clusters
Labels = None array of shape (n) prior Labels.

if None or inadequate a random initialization is performed.

maxiter=300 (int), the maximum number of iterations before convergence
delta: float, optional,

the relative increment in the results before declaring convergence.

verbose: verbosity mode, optional
ninit: int, optional, number of random initializations
Centers: array of shape (nbclusters, p),

the centroids of the resulting clusters

Labelsarray of size n, the discrete labels of the input items
J (float): the final value of the inertia criterion
nipy.algorithms.clustering.utils.voronoi(x, centers)

Assignment of data items to nearest cluster center

x array of shape (n,p)

n = number of items, p = data dimension

centers, array of shape (k, p) the cluster centers
z vector of shape(n), the resulting assignment