Smoothly Clipped Absolute Deviation Cox Regression

Smoothly Clipped Absolute Deviation (SCAD) Cox regression for high-dimensional survival data. SCAD provides automatic variable selection with oracle properties, avoiding over-penalization of large coefficients while maintaining sparsity for small coefficients. Particularly useful for genomics and high-dimensional clinical data where interpretable variable selection is crucial.

Usage

ncvregcox(
  data,
  time,
  event,
  covariates,
  penalty = "SCAD",
  cv_folds = 10,
  lambda_type = "min",
  alpha = 1,
  gamma = 3.7,
  standardize = TRUE,
  plot_path = TRUE,
  plot_cv = TRUE,
  variable_importance = TRUE
)

Arguments

data: the data as a data frame
time: survival time variable
event: event indicator (1=event, 0=censored)
covariates: predictor variables for high-dimensional analysis
penalty: Type of penalty function for variable selection
cv_folds: Number of folds for cross-validation
lambda_type: Lambda selection criterion
alpha: Mixing parameter for elastic net (1=lasso, 0=ridge)
gamma: SCAD gamma parameter or MCP gamma parameter
standardize: Standardize covariates before fitting
plot_path: Display coefficient paths plot
plot_cv: Display cross-validation error plot
variable_importance: Calculate and display variable importance metrics

Value

A results object containing:

`results$instructions`					a html
`results$model_summary`					a table
`results$selected_variables`					a table
`results$variable_importance`					a table
`results$cross_validation_results`					a table
`results$model_comparison`					a table
`results$convergence_info`					a table
`results$regularization_path`					an image
`results$cv_error_plot`					an image
`results$variable_selection_plot`					an image
`results$coefficient_comparison`					an image
`results$model_interpretation`					a html

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

results$model_summary$asDF

as.data.frame(results$model_summary)

Examples

ncvregcox(
    data = data,
    time = "time",
    event = "event",
    covariates = c("x1", "x2", "x3"),
    penalty = "SCAD",
    cv_folds = 10,
    lambda_type = "min"
)