popari.tl.compute_ari_scores

popari.tl.compute_ari_scores#

popari.tl.compute_ari_scores(dataset, labels, predictions, ari_key='ari')#

Compute adjusted Rand index (ARI) score between a set of ground truth labels and an unsupervised clustering.

Useful for assessing clustering validity. ARI score is computed per dataset.

Parameters:

  • dataset (PopariDataset) – dataset to process

  • labels (str) – the key in the .obs dataframe for the label data.

  • predictions (str) – the key in the .obs dataframe for the predictions data.

  • ari_key (str) – the key in the .uns dictionary where the ARI score will be stored.

Contents