drcme.post_gmm_merging.entropy_specific_merges

drcme.post_gmm_merging.entropy_specific_merges(tau, labels, K_bic, clusters_to_merge)[source]

Merge set of specified clusters by entropy criterion

Parameters
  • tau (array) – Cluster membership probabilities

  • labels (array) – Cluster assignments

  • K_bic (int) – Number of original clusters (before merging)

  • clusters_to_merge (array) – Set of clusters to merge

Returns

  • merge_info (dict) – Contains information about what was merged: entropies at each merge point (“entropies”), sequence of merges by indices as the matrix changes size (“merge_sequence”), sequence of merges by original column name (“merges_by_name”), original cluster number (“K_bic”), sequence of cluster numbers (“Ks”), number of cumulative points merged (“cumul_merges”)

  • new_labels (array) – Labels after merging

  • tau_merged (array) – Cluster membership probabilities after merging

  • merge_matrix (array) – Matrix of relationships between original clusters and merged clusters