Run a complete SOM defensibility workflow — run_som

The workflow preserves representation quality, partition stability, sample-level consensus and cross-model agreement as separate evidence streams. It does not combine them into a score or select a winning k.

Usage

run_som_workflow(
  data,
  spec,
  resamples = NULL,
  preprocess = som_preprocess(),
  k = spec$k,
  cross_models = c("kmeans", "ward"),
  cross_model_control = list(),
  consensus_method = "auto",
  max_coassignment_n = 5000L,
  keep_models = FALSE,
  fail_fast = FALSE,
  parallel = FALSE,
  max_pairwise_comparisons = 1000000L
)

Arguments

data: A som_data object.
spec: A som_spec object.
resamples: Optional som_resamples object. If NULL, a single full-data split is used; that run does not assess perturbation under resampling.
preprocess: One som_preprocess object or a named object per layer.
k: Candidate numbers of hard partitions.
cross_models: Controlled reference methods passed to fit_cross_models(). Use character() to skip cross-model fitting.
cross_model_control: Named list of optional kmeans_seeds, kmeans_nstart, kmeans_iter_max, gmm_model_names or gmm_seed settings forwarded to fit_cross_models(). Methods, candidate k, model retention and failure policy remain controlled by this workflow.
consensus_method: Consensus method passed to consensus_som().
max_coassignment_n: Dense co-assignment limit passed to consensus_som().
keep_models: Whether to retain fitted SOM and reference model objects.
fail_fast: Whether a model or consensus failure should stop the run.
parallel: Whether SOM ensemble members should use the current future plan; passed to fit_som_ensemble().
max_pairwise_comparisons: Pairwise partition-comparison budget passed to partition_som().

Value

A som_workflow object containing each evidence stream and explicit failure logs.

Examples

data <- simulate_som_scenario("clusters", n = 60, p = 4, seed = 6)
specification <- som_spec(c(3, 2), seeds = 1:2, rlen = 10, k = 2:3)
workflow <- run_som_workflow(
  data, specification, cross_models = c("kmeans", "ward")
)
summary(workflow)
#> <summary.som_workflow>
#>   SOM success: 2 / 2 
#>   consensus  : 2 computed; 0 not_computed
#>     - k=2: computed; complete=yes; assignment=100.0%; labels=100.0%; replicated=100.0%
#>     - k=3: computed; complete=yes; assignment=100.0%; labels=100.0%; replicated=100.0%
#>   reference methods:
#>     - kmeans: 2 expected, 2 succeeded, 0 failed, 0 warnings, 100.0% success
#>     - ward: 2 expected, 2 succeeded, 0 failed, 0 warnings, 100.0% success