Struct SelectFromModelExt 

pub struct SelectFromModelExt<F> { /* private fields */ }

An extended model-importance-based feature selector.

Like SelectFromModel but supports four threshold strategies and an optional max_features cap.

Examples

use ferrolearn_preprocess::select_from_model::{SelectFromModelExt, ThresholdStrategy};
use ferrolearn_core::traits::{Fit, Transform};
use ndarray::array;

let sel = SelectFromModelExt::<f64>::new(ThresholdStrategy::Mean, None);
let importances = array![0.1, 0.5, 0.4];
let fitted = sel.fit(&importances, &()).unwrap();
let x = array![[1.0, 2.0, 3.0], [4.0, 5.0, 6.0]];
let out = fitted.transform(&x).unwrap();
// Mean importance = (0.1+0.5+0.4)/3 ≈ 0.333; columns 1 and 2 kept
assert_eq!(out.ncols(), 2);

Implementations

impl<F: Float + Send + Sync + 'static> SelectFromModelExt<F>

pub fn new(threshold: ThresholdStrategy, max_features: Option<usize>) -> Self

Create a new SelectFromModelExt.

Parameters

• threshold — the strategy for computing the importance threshold.
• max_features — optional maximum number of features to retain.

pub fn threshold_strategy(&self) -> ThresholdStrategy

Return the threshold strategy.

pub fn max_features(&self) -> Option<usize>

Return the maximum number of features (if set).

Trait Implementations

impl<F: Clone> Clone for SelectFromModelExt<F>

fn clone(&self) -> SelectFromModelExt<F>

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl<F: Debug> Debug for SelectFromModelExt<F>

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<F: Float + Send + Sync + 'static> Default for SelectFromModelExt<F>

fn default() -> Self

Returns the “default value” for a type.

impl<F: Float + Send + Sync + 'static> Fit<ArrayBase<OwnedRepr<F>, Dim<[usize; 1]>>, ()> for SelectFromModelExt<F>

fn fit( &self, x: &Array1<F>, _y: &(), ) -> Result<FittedSelectFromModelExt<F>, FerroError>

Fit by computing the threshold from the given feature importances.

Parameters

• x — per-feature importance scores (one value per feature).
• _y — ignored (unsupervised).

Errors

• FerroError::InvalidParameter if the importance vector is empty, or if Percentile value is not in (0, 100].

type Fitted = FittedSelectFromModelExt<F>

The fitted model type returned by fit.

type Error = FerroError

The error type returned by fit.