test_k_means.py 文件源码

def test_k_means_non_collapsed():
    # Check k_means with a bad initialization does not yield a singleton
    # Starting with bad centers that are quickly ignored should not
    # result in a repositioning of the centers to the center of mass that
    # would lead to collapsed centers which in turns make the clustering
    # dependent of the numerical unstabilities.
    my_X = np.array([[1.1, 1.1], [0.9, 1.1], [1.1, 0.9], [0.9, 1.1]])
    array_init = np.array([[1.0, 1.0], [5.0, 5.0], [-5.0, -5.0]])
    km = KMeans(init=array_init, n_clusters=3, random_state=42, n_init=1)
    km.fit(my_X)

    # centers must not been collapsed
    assert_equal(len(np.unique(km.labels_)), 3)

    centers = km.cluster_centers_
    assert_true(np.linalg.norm(centers[0] - centers[1]) >= 0.1)
    assert_true(np.linalg.norm(centers[0] - centers[2]) >= 0.1)
    assert_true(np.linalg.norm(centers[1] - centers[2]) >= 0.1)