Struct SelectFromModel 

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

A feature selector driven by external feature-importance weights.

The importance vector is typically obtained from a fitted model (e.g. a decision-tree model’s feature_importances_ field). Features whose importance is strictly greater than or equal to the threshold are kept.

The default threshold is the mean importance of all features.

Examples

use ferrolearn_preprocess::feature_selection::SelectFromModel;
use ferrolearn_core::traits::Transform;
use ndarray::array;

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

Implementations

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

pub fn new_from_importances( importances: &Array1<F>, threshold: Option<F>, ) -> Result<Self, FerroError>

Create a SelectFromModel from a pre-computed importance vector.

Parameters

• importances — one importance weight per feature.
• threshold — optional explicit threshold; if None the mean importance is used.

Errors

Returns FerroError::InvalidParameter if importances is empty.

pub fn threshold(&self) -> F

Return the threshold used to select features.

pub fn importances(&self) -> &Array1<F>

Return the importance vector supplied at construction time.

pub fn selected_indices(&self) -> &[usize]

Return the indices of the selected columns.

Trait Implementations

impl<F: Clone> Clone for SelectFromModel<F>

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

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<F: Debug> Debug for SelectFromModel<F>

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

Formats the value using the given formatter.

impl<F: Float + Send + Sync + 'static> FittedPipelineTransformer<F> for SelectFromModel<F>

fn transform_pipeline(&self, x: &Array2<F>) -> Result<Array2<F>, FerroError>

Transform using the pipeline interface.

Errors

Propagates errors from Transform::transform.

impl<F: Float + Send + Sync + 'static> PipelineTransformer<F> for SelectFromModel<F>

fn fit_pipeline( &self, _x: &Array2<F>, _y: &Array1<F>, ) -> Result<Box<dyn FittedPipelineTransformer<F>>, FerroError>

Clone the selector and box it as a fitted pipeline transformer.

Errors

This implementation never fails.

impl<F: Float + Send + Sync + 'static> Transform<ArrayBase<OwnedRepr<F>, Dim<[usize; 2]>>> for SelectFromModel<F>

fn transform(&self, x: &Array2<F>) -> Result<Array2<F>, FerroError>

Return a matrix containing only the columns whose importance exceeds the threshold.

Errors

Returns FerroError::ShapeMismatch if the number of columns differs from the length of the importance vector supplied at construction.

type Output = ArrayBase<OwnedRepr<F>, Dim<[usize; 2]>>

The transformed output type.

type Error = FerroError

The error type returned by transform.