Exemplo n.º 1
0
def spectral_coclustering(A, use_k_means=False, k=3):

    D1 = np.sum(A, 1)
    D2 = np.sum(A, 0)


    D1_root = np.diag(np.abs((D1)**(-0.5)))
    D2_root = np.diag(np.abs((D2)**(-0.5)))

    An = dot(D1_root, dot(A, D2_root))

    #print An
    t1 = time()
    U, S, V = svd(An)
    V = np.transpose(V)
    #print U, S, V
    print "%s" % (time() - t1)

    n_eigen = np.ceil(np.log2(k))
    z = np.vstack((dot(D1_root, U[:, 1:n_eigen+1]), 
            dot(D2_root, V[:,1:n_eigen+1])))

    if use_k_means:
        _, labels, _ = k_means(z, k)

    else:
        t1 = time()
        labels = ap(z)
        print "%s" % (time() - t1)

    r_labels = labels[:A.shape[0]]
    c_labels = labels[A.shape[0]:]
    return r_labels, c_labels
Exemplo n.º 2
0
def spectral_coclustering(A, use_k_means=False, k=3):

    D1 = np.sum(A, 1)
    D2 = np.sum(A, 0)

    D1_root = np.diag(np.abs((D1)**(-0.5)))
    D2_root = np.diag(np.abs((D2)**(-0.5)))

    An = dot(D1_root, dot(A, D2_root))

    #print An
    t1 = time()
    U, S, V = svd(An)
    V = np.transpose(V)
    #print U, S, V
    print "%s" % (time() - t1)

    n_eigen = np.ceil(np.log2(k))
    z = np.vstack(
        (dot(D1_root, U[:, 1:n_eigen + 1]), dot(D2_root, V[:, 1:n_eigen + 1])))

    if use_k_means:
        _, labels, _ = k_means(z, k)

    else:
        t1 = time()
        labels = ap(z)
        print "%s" % (time() - t1)

    r_labels = labels[:A.shape[0]]
    c_labels = labels[A.shape[0]:]
    return r_labels, c_labels
Exemplo n.º 3
0
 def k_means_plus(*args, **kwargs):
     """k means with smart initialization.
    
     Notes
     -----
     This function requires scikits-learn
     
     See Also
     --------
     kmeans
     """
     return skcluster.k_means(*args, **kwargs)[1]
Exemplo n.º 4
0
 def k_means_plus(*args, **kwargs):
     """k means with smart initialization.
    
     Notes
     -----
     This function requires scikits-learn
     
     See Also
     --------
     kmeans
     """
     return skcluster.k_means(*args, **kwargs)[1]
Exemplo n.º 5
0
import numpy as np
from numpy.linalg import svd
from numpy import dot
from time import time
from scipy.cluster.hierarchy import dendrogram, linkage

#fast version
from scikits.learn.utils.extmath import fast_svd
from scikits.learn.cluster import k_means, affinity_propagation

if __name__ == "__main__":
    A = np.loadtxt('town.csv', delimiter=",")

    _, labels, _ = k_means(A, 3)
    print labels
Exemplo n.º 6
0
import numpy as np
from numpy.linalg import svd
from numpy import dot
from time import time
from scipy.cluster.hierarchy import dendrogram, linkage

#fast version
from scikits.learn.utils.extmath import fast_svd
from scikits.learn.cluster import k_means, affinity_propagation




if __name__ == "__main__":
    A = np.loadtxt('town.csv', delimiter=",")

    _, labels, _ = k_means(A, 3)
    print labels