Struct GaussianNb

pub struct GaussianNb<F: PartialEq, L: Eq + Hash> { /* private fields */ }

Fitted Gaussian Naive Bayes classifier.

See GaussianNbParams for more information on the hyper-parameters.

Model assumptions

The family of Naive Bayes classifiers assume independence between variables. They do not model moments between variables and lack therefore in modelling capability. The advantage is a linear fitting time with maximum-likelihood training in a closed form.

Model usage example

The example below creates a set of hyperparameters, and then uses it to fit a Gaussian Naive Bayes classifier on provided data.

use linfa_bayes::{GaussianNbParams, GaussianNbValidParams, Result};
use linfa::prelude::*;
use ndarray::array;

let x = array![
    [-2., -1.],
    [-1., -1.],
    [-1., -2.],
    [1., 1.],
    [1., 2.],
    [2., 1.]
];
let y = array![1, 1, 1, 2, 2, 2];
let ds = DatasetView::new(x.view(), y.view());

// create a new parameter set with variance smoothing equals `1e-5`
let unchecked_params = GaussianNbParams::new()
    .var_smoothing(1e-5);

// fit model with unchecked parameter set
let model = unchecked_params.fit(&ds)?;

// transform into a verified parameter set
let checked_params = unchecked_params.check()?;

// update model with the verified parameters, this only returns
// errors originating from the fitting process
let model = checked_params.fit_with(Some(model), &ds)?;

Implementations

impl<F: Float, L: Label> GaussianNb<F, L>

pub fn params() -> GaussianNbParams<F, L>

Construct a new set of hyperparameters

Trait Implementations

impl<F: Clone + PartialEq, L: Clone + Eq + Hash> Clone for GaussianNb<F, L>

fn clone(&self) -> GaussianNb<F, L>

Returns a copy of the value.

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

Performs copy-assignment from source.

impl<F: Debug + PartialEq, L: Debug + Eq + Hash> Debug for GaussianNb<F, L>

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

Formats the value using the given formatter.

impl<F, L> NaiveBayes<'_, F, L> for GaussianNb<F, L>

where F: Float, L: Label + Ord,

fn joint_log_likelihood(&self, x: ArrayView2<'_, F>) -> HashMap<&L, Array1<F>>

Compute the unnormalized posterior log probabilities. The result is returned as an HashMap indexing log probabilities for each samples (eg x rows) by classes (eg jll[class] -> (n_samples,) array)

fn predict_log_proba(&self, x: ArrayView2<'_, F>) -> (Array2<F>, Vec<&L>)

Compute log-probability estimates for each sample wrt classes. The columns corresponds to classes in sorted order returned as the second output.

fn predict_proba(&self, x: ArrayView2<'_, F>) -> (Array2<F>, Vec<&L>)

Compute probability estimates for each sample wrt classes. The columns corresponds to classes in sorted order returned as the second output.

impl<F: PartialEq + PartialEq, L: PartialEq + Eq + Hash> PartialEq for GaussianNb<F, L>

fn eq(&self, other: &GaussianNb<F, L>) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl<F: Float, L: Label, D> PredictInplace<ArrayBase<D, Dim<[usize; 2]>>, ArrayBase<OwnedRepr<L>, Dim<[usize; 1]>>> for GaussianNb<F, L>

where D: Data<Elem = F>,

fn predict_inplace(&self, x: &ArrayBase<D, Ix2>, y: &mut Array1<L>)

Predict something in place

fn default_target(&self, x: &ArrayBase<D, Ix2>) -> Array1<L>

Create targets that predict_inplace works with.

impl<F: PartialEq, L: Eq + Hash> StructuralPartialEq for GaussianNb<F, L>