Skip to main content

stability_selection

fdars_core::boosting_regression::stability

Function stability_selection 

Source 
pub fn stability_selection(
    data: &FdMatrix,
    predictors: &FdMatrix,
    argvals: &[f64],
    boost_config: &BoostingConfig,
    stab_config: &StabilityConfig,
) -> Result<StabilityResult, FdarError>
Expand description

FDboost-style stability selection.

Runs stab_config.n_resamples subsamples of size ⌊n/2⌋ (without replacement), fits boost_fosr on each subsample, and aggregates per-base-learner selection frequencies across resamples.

§Arguments

  • data — Functional response Y (n × m_t).
  • predictors — Scalar predictor matrix (n × p). One base-learner per column.
  • argvals — Response grid evaluation points (length m_t).
  • boost_config — BoostingConfig for the inner boost_fosr calls.
  • stab_config — StabilityConfig controlling resamples, threshold, and seed.

§Returns

StabilityResult with per-learner selection frequencies, the stable set at pi_thr, the PFER bound, and the number of resamples used.

§Errors

FdarError::InvalidDimension on shape problems (including too-small subsamples); FdarError::InvalidParameter on out-of-range config; propagates any FdarError from the inner boost_fosr fits.