scipy.cluster.hierarchy.

cut_tree#

scipy.cluster.hierarchy.cut_tree(Z, n_clusters=None, height=None)[source]#

Given a linkage matrix Z, return the cut tree.

Parameters:

Zscipy.cluster.linkage array: The linkage matrix.
n_clustersarray_like, optional: Number of clusters in the tree at the cut point.
heightarray_like, optional: The height at which to cut the tree. Only possible for ultrametric trees.

Returns:

cutreearray: An array indicating group membership at each agglomeration step. I.e., for a full cut tree, in the first column each data point is in its own cluster. At the next step, two nodes are merged. Finally, all singleton and non-singleton clusters are in one group. If n_clusters or height are given, the columns correspond to the columns of n_clusters or height.

Notes

cut_tree has experimental support for Python Array API Standard compatible backends in addition to NumPy. Please consider testing these features by setting an environment variable SCIPY_ARRAY_API=1 and providing CuPy, PyTorch, JAX, or Dask arrays as array arguments. The following combinations of backend and device (or other capability) are supported.

Library	CPU	GPU
NumPy	✅	n/a
CuPy	n/a	⛔
PyTorch	⛔	⛔
JAX	⛔	⛔
Dask	⛔	n/a

See Support for the array API standard for more information.

Examples

>>> from scipy import cluster
>>> import numpy as np
>>> from numpy.random import default_rng
>>> rng = default_rng()
>>> X = rng.random((50, 4))
>>> Z = cluster.hierarchy.ward(X)
>>> cutree = cluster.hierarchy.cut_tree(Z, n_clusters=[5, 10])
>>> cutree[:10]
array([[0, 0],
       [1, 1],
       [2, 2],
       [3, 3],
       [3, 4],
       [2, 2],
       [0, 0],
       [1, 5],
       [3, 6],
       [4, 7]])  # random