evaluator_clustering#

Source code: sensai/evaluation/evaluator_clustering.py

class ClusteringModelEvaluator(*args, **kwds)[source]#

Bases: MetricsDictProvider, Generic[TClusteringEvalStats], ABC

abstract eval_model(model: EuclideanClusterer, **kwargs) → TClusteringEvalStats[source]#
class ClusteringModelUnsupervisedEvaluator(datapoints)[source]#

Bases: ClusteringModelEvaluator[ClusteringUnsupervisedEvalStats]

eval_model(model: EuclideanClusterer, fit=True)[source]#

Retrieve evaluation statistics holder for the clustering model

Parameters:

  • model –

  • fit – whether to fit on the evaluator’s data before retrieving statistics. Set this to False if the model you wish to evaluate was already fitted on the desired dataset

Returns:

instance of ClusteringUnsupervisedEvalStats that can be used for calculating various evaluation metrics

class ClusteringModelSupervisedEvaluator(datapoints, true_labels: Sequence[int], noise_label=- 1)[source]#

Bases: ClusteringModelEvaluator[ClusteringSupervisedEvalStats]

Parameters:

  • datapoints –

  • true_labels – labels of the true clusters, including the noise clusters.

  • noise_label – label of the noise cluster (if there is one) in the true labels

eval_model(model: EuclideanClusterer, fit=True)[source]#

Retrieve evaluation statistics holder for the clustering model

Parameters:

  • model –

  • fit – whether to fit on the evaluator’s data before retrieving statistics. Set this to False if the model you wish to evaluate was already fitted on the desired dataset

Returns:

instance of ClusteringSupervisedEvalStats that can be used for calculating various evaluation metrics