HierarchicalClustering

class HierarchicalClustering(distance: etna.clustering.distances.base.Distance)

Bases: etna.clustering.base.Clustering

Base class for hierarchical clustering.

Init HierarchicalClustering.

Inherited-members

Parameters

distance (etna.clustering.distances.base.Distance) –

Methods

build_clustering_algo([n_clusters, linkage])	Build clustering algo (see sklearn.cluster.AgglomerativeClustering) with given params.
build_distance_matrix(ts)	Compute distance matrix with given ts and distance.
fit_predict()	Fit clustering algorithm and predict clusters according to distance matrix build.
get_centroids(**averaging_kwargs)	Get centroids of clusters.
set_params(**params)	Return new object instance with modified parameters.
to_dict()	Collect all information about etna object in dict.

build_clustering_algo(n_clusters: int = 30, linkage: Union[str, etna.clustering.hierarchical.base.ClusteringLinkageMode] = ClusteringLinkageMode.average, **clustering_algo_params)

Build clustering algo (see sklearn.cluster.AgglomerativeClustering) with given params.

Parameters

• n_clusters (int) – number of clusters to build

• linkage (Union[str, etna.clustering.hierarchical.base.ClusteringLinkageMode]) – rule for distance computation for new clusters, allowed “ward”, “single”, “average”, “maximum”, “complete”

Notes

Note that it will reset previous results of clustering in case of reinit algo.

build_distance_matrix(ts: etna.datasets.tsdataset.TSDataset)

Compute distance matrix with given ts and distance.

Parameters

• ts (etna.datasets.tsdataset.TSDataset) – TSDataset with series to build distance matrix

• distance – instance if distance to compute matrix

fit_predict() → Dict[str, int]

Fit clustering algorithm and predict clusters according to distance matrix build.

Returns

dict in format {segment: cluster}

Return type

Dict[str, int]

get_centroids(**averaging_kwargs) → pandas.core.frame.DataFrame

Get centroids of clusters.

Returns

dataframe with centroids

Return type

pd.DataFrame