Struct SelectFromModel 

pub struct SelectFromModel {
    pub threshold: Option<f64>,
    pub max_features: Option<usize>,
}

Parameters for SelectFromModel feature selector (unfitted state).

Selects features based on a pre-computed importance vector (e.g., from a fitted RandomForestClassifier’s feature_importances()).

Features can be selected in two ways:

• Threshold: keep all features with importance >= threshold.
• Max features: keep the top max_features features by importance.

If both are specified, threshold is applied first and then the result is capped to max_features.

Since this selector does not learn from raw data (it uses pre-computed importances), it exposes a custom fit method instead of implementing the standard FitUnsupervised trait.

Example

use anofox_ml_preprocessing::SelectFromModel;
use anofox_ml_core::Transform;
use ndarray::array;

let importances = array![0.05, 0.40, 0.10, 0.45];

// Select features with importance >= 0.20
let selector = SelectFromModel::new().with_threshold(0.20);
let fitted = selector.fit(&importances).unwrap();

assert_eq!(fitted.selected_indices(), &[1, 3]);

let x = array![
    [1.0, 2.0, 3.0, 4.0],
    [5.0, 6.0, 7.0, 8.0],
];
let x_selected = fitted.transform(&x).unwrap();
assert_eq!(x_selected.ncols(), 2);

Fields

threshold: Option<f64>

Features with importance >= threshold are selected. If None, no threshold filtering is applied.

max_features: Option<usize>

Maximum number of features to select (top by importance). If None, no cap is applied.

Implementations

impl SelectFromModel

pub fn new() -> Self

Create a new SelectFromModel with no threshold and no feature cap.

At least one of threshold or max_features must be set before calling fit.

pub fn with_threshold(self, threshold: f64) -> Self

Set the minimum importance threshold.

pub fn with_max_features(self, max_features: usize) -> Self

Set the maximum number of features to select.

pub fn fit(&self, importances: &Array1<f64>) -> Result<FittedSelectFromModel>

Fit the selector on a pre-computed feature importance vector.

Returns an error if the importance vector is empty or if neither threshold nor max_features is set.

Trait Implementations

impl Clone for SelectFromModel

fn clone(&self) -> SelectFromModel

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for SelectFromModel

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

Formats the value using the given formatter.

impl Default for SelectFromModel

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for SelectFromModel

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for SelectFromModel

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.