Spike-and-Slab Variable Selection

Advanced Bayesian variable selection using spike-and-slab priors for high-dimensional data analysis. Supports regression, classification, and survival analysis with automatic variable selection, model averaging, and uncertainty quantification for clinical biomarker discovery.

Usage

spikeslabpriors(
  data,
  outcome,
  predictors,
  outcomeLevel,
  time_variable = NULL,
  status_variable = NULL,
  model_type = "regression",
  spike_slab_type = "binary",
  prior_inclusion_prob = 0.5,
  expected_model_size = 5,
  spike_variance = 0.001,
  slab_variance = 1,
  hyperprior_type = "empirical_bayes",
  hyperprior_alpha = 1,
  hyperprior_beta = 1,
  selection_criterion = "median_probability",
  inclusion_threshold = 0.5,
  bayes_factor_threshold = 3,
  mcmc_samples = 10000,
  mcmc_burnin = 5000,
  mcmc_thin = 1,
  mcmc_chains = 3,
  model_averaging = TRUE,
  prediction_method = "bma",
  cross_validation = FALSE,
  cv_folds = 10,
  standardize_predictors = TRUE,
  center_predictors = TRUE,
  max_model_size = 50,
  dimension_reduction = "none",
  prescreening_threshold = 0.1,
  show_variable_selection = TRUE,
  show_inclusion_probabilities = TRUE,
  show_model_probabilities = TRUE,
  show_coefficient_estimates = TRUE,
  show_prediction_performance = TRUE,
  show_convergence_diagnostics = TRUE,
  show_variable_importance = TRUE,
  show_model_comparison = FALSE,
  show_interpretation = TRUE,
  interaction_terms = FALSE,
  max_interactions = 2,
  heredity_constraint = "weak",
  group_variables = FALSE,
  group_structure = NULL,
  clinical_context = "biomarker_discovery",
  confidence_level = 0.95,
  set_seed = TRUE,
  seed_value = 42,
  parallel_processing = TRUE,
  n_cores = 2
)

Arguments

data

The data as a data frame for spike-and-slab variable selection.

outcome

.

predictors

.

outcomeLevel

.

time_variable

For survival analysis

status_variable

For survival analysis

model_type

.

spike_slab_type

.

prior_inclusion_prob

Prior probability that each variable is included

expected_model_size

Expected number of active predictors

spike_variance

Variance for spike component (near-zero effects)

slab_variance

Variance for slab component (non-zero effects)

hyperprior_type

.

hyperprior_alpha

.

hyperprior_beta

.

selection_criterion

.

inclusion_threshold

.

bayes_factor_threshold

.

mcmc_samples

.

mcmc_burnin

.

mcmc_thin

.

mcmc_chains

.

model_averaging

.

prediction_method

.

cross_validation

.

cv_folds

.

standardize_predictors

.

center_predictors

.

max_model_size

Maximum number of variables to include

dimension_reduction

.

prescreening_threshold

.

show_variable_selection

.

show_inclusion_probabilities

.

show_model_probabilities

.

show_coefficient_estimates

.

show_prediction_performance

.

show_convergence_diagnostics

.

show_variable_importance

.

show_model_comparison

.

show_interpretation

.

interaction_terms

.

max_interactions

.

heredity_constraint

.

group_variables

.

group_structure

.

clinical_context

.

confidence_level

.

set_seed

.

seed_value

.

parallel_processing

.

n_cores

.

Value

A results object containing:

results$variableSelection					a table
results$inclusionProbabilities					a table
results$modelProbabilities					a table
results$coefficientEstimates					a table
results$predictionPerformance					a table
results$convergenceDiagnostics					a table
results$variableImportance					a table
results$modelComparison					a table
results$dimensionSummary					a table
results$groupSelection					a table
results$clinicalInterpretation					a html
results$methodsExplanation					a html
results$inclusionProbabilityPlot					an image
results$modelSizePlot					an image
results$coefficientPlot					an image
results$variableImportancePlot					an image
results$convergencePlots					an image
results$posteriorDistributionPlots					an image
results$modelComparisonPlot					an image
results$predictionPerformancePlot					an image

Tables can be converted to data frames with asDF or as.data.frame. For example:

results$variableSelection$asDF

as.data.frame(results$variableSelection)

Examples

# Spike-and-slab variable selection
spikeslabpriors(
    data = biomarker_data,
    outcome = "clinical_response",
    predictors = c("gene1", "gene2", "gene3", ..., "gene1000"),
    model_type = "regression",
    prior_inclusion_prob = 0.1
)