Skip to contents

Fit HDBSCAN clusters on a frozen self-organizing map

Source: R/SOMHDBSCAN.R
CreateClusterModel_SOM_HDBSCAN.Rd

Trains the same frozen SOM used by CreateClusterModel_SOM_MClust(), then applies HDBSCAN to its node codebook. Participants inherit the cluster (or noise) label of their best-matching node. This is a node-based phenotype model: projected participants are mapped to the original nodes and are not refit with HDBSCAN.

Usage

CreateClusterModel_SOM_HDBSCAN(
  data,
  variables = NULL,
  method = c("exploratory", "finalize"),
  minPts_range = 2:10,
  cluster_selection_epsilon_range = c(0, 0.05, 0.1),
  final_minPts = NULL,
  final_cluster_selection_epsilon = NULL,
  ClusterVariableName = "Cluster",
  seed_som = 934521L,
  seed_hdbscan = 93421L,
  stability_resamples = 0L,
  stability_seed = seed_hdbscan + 1L,
  stability_progress = FALSE,
  ...
)

Arguments

data

Data frame used to train the SOM and node-level HDBSCAN model.

variables

Numeric variables used for SOM training.

method

Either "exploratory" or "finalize".

minPts_range

Candidate HDBSCAN minimum-point settings for SOM nodes.

cluster_selection_epsilon_range

Candidate HDBSCAN extraction epsilon settings.

final_minPts

Optional finalized HDBSCAN minimum-point setting.

final_cluster_selection_epsilon

Optional finalized extraction epsilon.

ClusterVariableName

Name of the appended cluster column.

seed_som

Seed used for SOM training.

seed_hdbscan

Seed retained in the model specification.

stability_resamples

Number of 80% participant subsample refits used for final-model stability. Subsamples are drawn without replacement.

stability_seed

Seed controlling participant subsampling.

stability_progress

Whether to print subsample progress messages.

...

Additional arguments passed to CreateClusterModel_SOM_MClust().

Value

A Pipeline_SOM_HDBSCAN object containing frozen SOM and HDBSCAN node models, assignments, diagnostics, and a projection specification. Persistence and minimum node-cluster size are higher-is-better, noise proportion is lower-is-better, and extracted class count is data-derived.

Stability output

Stability assesses internal reproducibility by full-pipeline 80% participant subsampling without replacement. For each replicate, 80% of complete participants are selected once, all preprocessing and any reduction (PCA, MCA, or SOM) are refit, the selected clustering method is refit, and the original complete training participants are projected into that subsample fit for comparison with the original fitted partition. It is an internal sensitivity analysis, not independent-cohort validation.

Stability$settings records the analysis provenance:

  • resamples: requested number of 80% subsample refits.

  • seed: seed used to select subsamples.

  • refit_scope: always "full_pipeline", meaning preprocessing, reduction where applicable, and clustering were all refit.

  • resample_type: "subsample_without_replacement" for the primary stability analysis.

  • resample_fraction: the retained participant fraction, 0.80.

  • coassignment_limit: maximum number of complete training participants (2,000) for which the full pairwise co-assignment matrix is calculated.

  • noise_policy: whether the method has noise labels. For ordinary methods it is "all clusters included"; HDBSCAN variants retain noise in global partition metrics but exclude it from per-cluster inclusion and co-assignment summaries.

Stability$replicates has one row per requested refit. Model and Classes identify the selected candidate (for HDBSCAN, Classes is the data-derived extracted count); Replicate is its sequence number; Status is "success" or a failure status; and Error contains the error message for an unsuccessful refit. Successful rows contain these partition metrics:

  • ARI: adjusted Rand index, agreement corrected for chance; higher is better and can be negative when agreement is worse than chance.

  • VI: variation of information, the information lost or gained when changing partitions; lower is better and zero is identical.

  • NMI: normalized mutual information; higher is better and one is identical.

  • FowlkesMallows: pairwise clustering agreement; higher is better and one is identical.

Stability$cluster_recovery has one row for each reference Cluster in each successful Replicate. Jaccard is the recovery of that reference cluster after matching it to the subsample cluster with the largest Jaccard overlap; higher is better and one is exact recovery. Model and Classes again identify the fitted candidate.

Stability$summary combines successful subsample refits: StabilitySuccessRate is successful refits divided by requested refits, an operational reliability measure that does not enter the reproducibility score. StabilityARI_Mean and StabilityARI_P05 are the mean and fifth percentile of ARI. StabilityJaccard_Mean and StabilityJaccard_Min are respectively the mean and minimum label-matched Jaccard recovery. ReproducibilityScore is the mean of the finite StabilityARI_Mean and StabilityJaccard_Mean values only; it does not include success rate, VI, NMI, or Fowlkes–Mallows.

Stability$failures repeats the replicate columns for unsuccessful refits, making fit failures auditable without mixing them with successful metrics.

Stability$participant_inclusion is one row per complete reference participant. RowIndex identifies its original row position, Cluster its reference assignment, SuccessfulRefits the number of usable refits, and InclusionProbability the proportion of those refits in which the participant returned to that cluster's label-matched subsample cluster. Model and Classes identify the candidate. Higher inclusion is better.

Stability$cluster_inclusion summarizes inclusion within each reference Cluster: MeanInclusion, P05Inclusion, and MinInclusion are the mean, fifth percentile, and minimum participant inclusion probabilities; Model and Classes identify the candidate. Higher values indicate that all, not only the average, of a cluster is recovered consistently.

Stability$coassignment is available only when the complete training cohort has at most coassignment_limit participants. Each candidate entry has a status of "available", "skipped", or "not_available"; reason explains a non-available result; matrix is the pairwise probability that two complete reference participants are assigned together across successful subsample refits; and row_ids maps matrix rows and columns to original training-row positions. Higher matrix values mean more consistent pairwise co-membership. Where more than one candidate is summarized, entries are named by its Model_Classes key. The matrix is diagnostic only and is never used for selection.

Stability$plots contains cluster_recovery (per-cluster Jaccard), partition_metrics (ARI, VI, NMI, and Fowlkes–Mallows distributions), and cluster_inclusion; it also contains a co-assignment heatmap when the matrix is available. These diagnostics complement rather than replace ARI and Jaccard: none can turn a poorly reproducible cluster into a stable phenotype.

Metric sources: Hubert and Arabie (1985) define ARI; Jaccard (1901) defines the overlap coefficient; Meila (2005) defines VI; Strehl and Ghosh (2002) describe NMI for partition comparison; Fowlkes and Mallows (1983) define their pairwise index; and Monti et al. (2003) describe resampling-based consensus co-assignment.