Struct GaussianProcessRegressor 

pub struct GaussianProcessRegressor<K: Kernel> { /* private fields */ }

Gaussian Process Regressor with scikit-learn compatible API

Examples

use scirs2_stats::gaussian_process::{GaussianProcessRegressor, SquaredExponential};
use scirs2_core::ndarray::{array, Array2};

let kernel = SquaredExponential::default();
let mut gpr = GaussianProcessRegressor::new(kernel);

let x_train = Array2::from_shape_vec((3, 1), vec![0.0, 1.0, 2.0]).expect("Operation failed");
let y_train = array![0.0, 1.0, 0.5];

gpr.fit(&x_train, &y_train).expect("Operation failed");

let x_test = Array2::from_shape_vec((1, 1), vec![1.5]).expect("Operation failed");
let predictions = gpr.predict(&x_test).expect("Operation failed");

Implementations

impl<K: Kernel> GaussianProcessRegressor<K>

pub fn new(kernel: K) -> Self

Create a new Gaussian Process Regressor

Arguments

• kernel - The covariance kernel

Returns

A new GaussianProcessRegressor with default settings

pub fn with_options(kernel: K, alpha: f64, normalize_y: bool) -> Self

Create a new GP Regressor with custom options

Arguments

• kernel - The covariance kernel
• alpha - Noise level / regularization parameter
• normalize_y - Whether to normalize target values

pub fn fit(&mut self, x: &Array2<f64>, y: &Array1<f64>) -> StatsResult<()>

Fit the Gaussian Process model

Arguments

• x - Training features (n_samples, n_features)
• y - Training targets (n_samples,)

Returns

Result indicating success or failure

pub fn predict(&self, x: &Array2<f64>) -> StatsResult<Array1<f64>>

Predict mean values

Arguments

• x - Test features (n_samples, n_features)

Returns

Predicted mean values

pub fn predict_with_std( &self, x: &Array2<f64>, ) -> StatsResult<(Array1<f64>, Array1<f64>)>

Predict with uncertainty estimates

Arguments

• x - Test features (n_samples, n_features)
• return_std - Whether to return standard deviation

Returns

(predictions, standard_deviations) if return_std is true

pub fn kernel(&self) -> &K

Get the kernel

pub fn kernel_mut(&mut self) -> &mut K

Get the kernel (mutable)

pub fn log_marginal_likelihood(&self) -> StatsResult<f64>

Compute log marginal likelihood

pub fn n_train_samples(&self) -> usize

Get number of training samples

pub fn score(&self, x: &Array2<f64>, y: &Array1<f64>) -> StatsResult<f64>

Score the model using R² metric

Arguments

• x - Test features
• y - True test targets

Returns

R² score