pycvi.cluster.compute_centers

pycvi.cluster.compute_centers(X: numpy.ndarray, clusters: List[List[int]] = [], keepdims: bool = False, avg_kwargs: dict = {}) → List[numpy.ndarray]

Compute the centers of all clusters.

In the case of static data

For non time-series data, this is simply the average of all datapoints in the given cluster using the usual mean function, and more precisely, calling numpy.mean.

In the case of time series data

For time-series data the cluster center is by default defined as the DBA (DTW barycentric average) as defined by Petitjean et al [DBA]. In this case, additional parameters can be passed in avg_kwargs, as described in aeon.clustering.averaging.elastic_barycenter_average. By default, uses { "distance": "msm", "init_barycenter": "medoids", "method": "petitjean", "random_state" : 221} .

For more information about the importance of using an elastic average instead of the euclidean mean for time series data, see our example Computing cluster centers

See pycvi.config.default_ts_average_kwargs() for more information about default averaging kwargs used in PyCVI.

[DBA]

F. Petitjean, A. Ketterlin, and P. Gan carski, “A global averaging method for dynamic time warping, with applications to clustering,” Pattern Recognition, vol. 44, pp. 678–693, Mar. 2011.

param X:

The original data.

type X:

np.ndarray, shape (N, d*w_t) or (N, w_t, d)

param clusters:

A list of clusters with indices.

type clusters:

List[List[int]]

param keepdims:

Whether to keep the dimension N of the input cluster, by default False.

type keepdims:

bool, optional

param avg_kwargs:

Keyword arguments for the average function. See pycvi.cluster.compute_center() and func:pycvi.cluster.compute_centers for more information.

type avg_kwargs:

dict, optional

returns:

A list of cluster centers. For each center:

• If keepdims=True then the shape is (1, d*w_t) or (1, w_t, d) if ts_dist=True.

• If keepdims=False then the shape is (d*w_t) or (w_t, d) if ts_dist=True.

rtype:

List[np.ndarray]