Struct TargetEncoder 

pub struct TargetEncoder { /* private fields */ }

TargetEncoder for supervised categorical encoding

This encoder transforms categorical features using the target variable values, encoding each category with a statistic (mean, median, etc.) of the target values for that category. This is useful for high-cardinality categorical features.

Examples

use ndarray::Array2;
use scirs2_transform::encoding::TargetEncoder;

let x = Array2::from_shape_vec((6, 1), vec![0.0, 1.0, 2.0, 0.0, 1.0, 2.0]).unwrap();
let y = vec![1.0, 2.0, 3.0, 1.5, 2.5, 3.5];

let mut encoder = TargetEncoder::new("mean", 1.0, 0.0).unwrap();
let encoded = encoder.fit_transform(&x, &y).unwrap();

Implementations

impl TargetEncoder

pub fn new(_strategy: &str, smoothing: f64, globalstat: f64) -> Result<Self>

Creates a new TargetEncoder

Arguments

• strategy - Encoding strategy (‘mean’, ‘median’, ‘count’, ‘sum’)
• smoothing - Smoothing parameter (0.0 = no smoothing, higher = more smoothing)
• globalstat - Global statistic fallback for unknown categories

Returns

• A new TargetEncoder instance

pub fn with_mean(smoothing: f64) -> Self

Creates a TargetEncoder with mean strategy and default smoothing

pub fn with_median(smoothing: f64) -> Self

Creates a TargetEncoder with median strategy

pub fn fit<S>(&mut self, x: &ArrayBase<S, Ix2>, y: &[f64]) -> Result<()>

where S: Data, S::Elem: Float + NumCast,

Fits the TargetEncoder to the input data and target values

Arguments

• x - The input categorical data, shape (n_samples, n_features)
• y - The target values, length n_samples

Returns

• Result<()> - Ok if successful, Err otherwise

pub fn transform<S>(&self, x: &ArrayBase<S, Ix2>) -> Result<Array2<f64>>

where S: Data, S::Elem: Float + NumCast,

Transforms the input data using the fitted TargetEncoder

Arguments

• x - The input categorical data, shape (n_samples, n_features)

Returns

• Result<Array2<f64>> - The target-encoded data

pub fn fit_transform<S>( &mut self, x: &ArrayBase<S, Ix2>, y: &[f64], ) -> Result<Array2<f64>>

where S: Data, S::Elem: Float + NumCast,

Fits the TargetEncoder and transforms the data in one step

Arguments

• x - The input categorical data, shape (n_samples, n_features)
• y - The target values, length n_samples

Returns

• Result<Array2<f64>> - The target-encoded data

pub fn encodings(&self) -> Option<&Vec<HashMap<u64, f64>>>

Returns the learned encodings for each feature

Returns

• Option<&Vec<HashMap<u64, f64>>> - The category encodings for each feature

pub fn is_fitted(&self) -> bool

Returns whether the encoder has been fitted

pub fn global_mean(&self) -> f64

Returns the global mean computed during fitting

pub fn fit_transform_cv<S>( &mut self, x: &ArrayBase<S, Ix2>, y: &[f64], cv_folds: usize, ) -> Result<Array2<f64>>

where S: Data, S::Elem: Float + NumCast,

Applies cross-validation target encoding to prevent overfitting

This method performs k-fold cross-validation to compute target encodings, which helps prevent overfitting when the same data is used for both fitting and transforming.

Arguments

• x - The input categorical data, shape (n_samples, n_features)
• y - The target values, length n_samples
• cv_folds - Number of cross-validation folds (default: 5)

Returns

• Result<Array2<f64>> - The cross-validated target-encoded data

Trait Implementations

impl Clone for TargetEncoder

fn clone(&self) -> TargetEncoder

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for TargetEncoder

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

Formats the value using the given formatter.