test_cluster.py 文件源码

def test_denrogram_children():
    # temporary solution for
    # https://stackoverflow.com/questions/40239956/node-indexing-in-hierarachical-clustering-dendrograms
    import numpy as np
    from scipy.cluster.hierarchy import dendrogram, linkage
    from freediscovery.cluster import _DendrogramChildren

    # generate two clusters: a with 10 points, b with 5:
    np.random.seed(1)
    a = np.random.multivariate_normal([10, 0], [[3, 1], [1, 4]],
                                      size=[10, ])
    b = np.random.multivariate_normal([0, 20], [[3, 1], [1, 4]],
                                      size=[5, ])
    X = np.concatenate((a, b),)
    Z = linkage(X, 'ward')
    # make distances between pairs of children uniform
    # (re-scales the horizontal (distance) axis when plotting)
    Z[:, 2] = np.arange(Z.shape[0])+1

    ddata = dendrogram(Z, no_plot=True)
    dc = _DendrogramChildren(ddata)
    idx = 0
    # check that we can compute children for all nodes
    for i, d, c in zip(ddata['icoord'], ddata['dcoord'], ddata['color_list']):
        node_children = dc.query(idx)
        idx += 1
    # last level node should encompass all samples
    assert len(node_children) == X.shape[0]
    assert_allclose(sorted(node_children), np.arange(X.shape[0]))