compounds.py 文件源码

def cluster(matrix):
    whitened = whiten(matrix.todense())

    # for x in range(25, 40):
    #     means, distortion = kmeans(whitened, x)
    #     print distortion

    means, distortion = kmeans(whitened, 30)

    # pickle.dump(means, open('30means-' + sys.argv[1] + '.pkl', 'wb'))

    return means, distortion