Struct SGDClassifier 

pub struct SGDClassifier<F> {
    pub loss: ClassifierLoss,
    pub learning_rate: LearningRateSchedule<F>,
    pub eta0: F,
    pub alpha: F,
    pub max_iter: usize,
    pub tol: F,
    pub random_state: Option<u64>,
    pub power_t: F,
}

Stochastic Gradient Descent classifier.

Supports binary classification via a decision boundary and multiclass classification via one-vs-all decomposition.

Type Parameters

• F: The floating-point type (f32 or f64).

Examples

use ferrolearn_linear::sgd::SGDClassifier;
use ferrolearn_core::{Fit, Predict};
use ndarray::{array, Array2};

let x = Array2::from_shape_vec((6, 2), vec![
    1.0, 2.0, 2.0, 3.0, 3.0, 1.0,
    8.0, 7.0, 9.0, 8.0, 7.0, 9.0,
]).unwrap();
let y = array![0, 0, 0, 1, 1, 1];

let clf = SGDClassifier::<f64>::new();
let fitted = clf.fit(&x, &y).unwrap();
let preds = fitted.predict(&x).unwrap();

Fields

loss: ClassifierLoss

The loss function to use.

learning_rate: LearningRateSchedule<F>

The learning rate schedule.

eta0: F

Initial learning rate.

alpha: F

L2 regularization strength.

max_iter: usize

Maximum number of passes over the training data.

tol: F

Convergence tolerance. Training stops when the loss improvement is below this threshold.

random_state: Option<u64>

Optional random seed for sample shuffling.

power_t: F

Power parameter for inverse scaling schedule.

Implementations

impl<F: Float> SGDClassifier<F>

pub fn new() -> Self

Create a new SGDClassifier with default settings.

Defaults: loss = Hinge, learning_rate = InvScaling, eta0 = 0.01, alpha = 0.0001, max_iter = 1000, tol = 1e-3, power_t = 0.25.

pub fn with_loss(self, loss: ClassifierLoss) -> Self

Set the loss function.

pub fn with_learning_rate(self, lr: LearningRateSchedule<F>) -> Self

Set the learning rate schedule.

pub fn with_eta0(self, eta0: F) -> Self

Set the initial learning rate.

pub fn with_alpha(self, alpha: F) -> Self

Set the L2 regularization strength.

pub fn with_max_iter(self, max_iter: usize) -> Self

Set the maximum number of epochs.

pub fn with_tol(self, tol: F) -> Self

Set the convergence tolerance.

pub fn with_random_state(self, seed: u64) -> Self

Set the random seed for reproducibility.

pub fn with_power_t(self, power_t: F) -> Self

Set the power parameter for inverse scaling.

Trait Implementations

impl<F: Clone> Clone for SGDClassifier<F>

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

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<F: Debug> Debug for SGDClassifier<F>

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

Formats the value using the given formatter.

impl<F: Float> Default for SGDClassifier<F>

fn default() -> Self

Returns the “default value” for a type.

impl<F: Float + Send + Sync + ScalarOperand + 'static> Fit<ArrayBase<OwnedRepr<F>, Dim<[usize; 2]>>, ArrayBase<OwnedRepr<usize>, Dim<[usize; 1]>>> for SGDClassifier<F>

fn fit( &self, x: &Array2<F>, y: &Array1<usize>, ) -> Result<FittedSGDClassifier<F>, FerroError>

Fit the SGD classifier on the given data.

Errors

Returns FerroError::ShapeMismatch if x and y have mismatched sample counts. Returns FerroError::InsufficientSamples if fewer than 2 classes are present. Returns FerroError::InvalidParameter if eta0 or alpha are not positive.

type Fitted = FittedSGDClassifier<F>

The fitted model type returned by fit.

type Error = FerroError

The error type returned by fit.

impl<F: Float + Send + Sync + ScalarOperand + 'static> PartialFit<ArrayBase<OwnedRepr<F>, Dim<[usize; 2]>>, ArrayBase<OwnedRepr<usize>, Dim<[usize; 1]>>> for SGDClassifier<F>

fn partial_fit( self, x: &Array2<F>, y: &Array1<usize>, ) -> Result<FittedSGDClassifier<F>, FerroError>

Initial call to partial_fit on an unfitted classifier.

Equivalent to fit but with a single epoch, enabling subsequent incremental calls.

Errors

Same as Fit::fit.

type FitResult = FittedSGDClassifier<F>

The result type returned by partial_fit.

type Error = FerroError

The error type returned by partial_fit.

impl<F> PipelineEstimator<F> for SGDClassifier<F>

where F: Float + ToPrimitive + FromPrimitive + ScalarOperand + Send + Sync + 'static,

fn fit_pipeline( &self, x: &Array2<F>, y: &Array1<F>, ) -> Result<Box<dyn FittedPipelineEstimator<F>>, FerroError>

Fit this estimator on the given data.