Struct linfa_ftrl::Ftrl

pub struct Ftrl<F: Float> { /* private fields */ }

Implementations

impl<F: Float> Ftrl<F>

View the fitted parameters and make predictions with a fitted follow the regularized leader -proximal, model

pub fn z(&self) -> &Array1<F>

Get Z values

pub fn n(&self) -> &Array1<F>

Get N values

pub fn alpha(&self) -> F

Get the alpha parameter

pub fn beta(&self) -> F

Get the beta parameter

pub fn l1_ratio(&self) -> F

Get the L1 regularization value

pub fn l2_ratio(&self) -> F

Get the L2 regularization value

pub fn get_weights(&self) -> Array1<F>

Calculate weights for model prediction

pub fn update<D: Data<Elem = F>, T: AsSingleTargets<Elem = bool>>(
    &mut self,
    dataset: &DatasetBase<ArrayBase<D, Ix2>, T>,
    probabilities: ArrayView1<'_, Pr>
)

Update method of the model hyperparameters in case of async mode. To use this method, we must store probabilities and features for the row, and update once the result (bool) is observed.

impl<F: Float> Ftrl<F>

pub fn params() -> FtrlParams<F, Xoshiro256Plus>

Create a default parameter set for construction of Follow The Regularized Leader - proximal model The description can be found here

It requires data preprocessing done in the separate step. Create default hyperparameters. Random number generator will default to rand_xoshiro::Xoshiro256Plus

pub fn params_with_rng<R: Rng>(rng: R) -> FtrlParams<F, R>

Create default hyperparameters with custom random number generator

pub fn new<R: Rng + Clone>(
    params: FtrlValidParams<F, R>,
    nfeatures: usize
) -> Ftrl<F>

Create a new model with given parameters, number of features and custom random number generator

Trait Implementations

impl<F: Clone + Float> Clone for Ftrl<F>

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

Returns a copy of the value.

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

Performs copy-assignment from source.

impl<F: Debug + Float> Debug for Ftrl<F>

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

Formats the value using the given formatter.

impl<'de, F: Float> Deserialize<'de> for Ftrl<F> where
    F: Deserialize<'de>, 

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error> where
    __D: Deserializer<'de>, 

Deserialize this value from the given Serde deserializer.

impl<F: Float, D: Data<Elem = F>> PredictInplace<ArrayBase<D, Dim<[usize; 2]>>, ArrayBase<OwnedRepr<Pr>, Dim<[usize; 1]>>> for Ftrl<F>

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

Given an input matrix X, with shape (n_samples, n_features), predict returns the target variable according to the parameters learned from the training data distribution.

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

Create targets that predict_inplace works with.

impl<F: Float> Serialize for Ftrl<F> where
    F: Serialize, 

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

Serialize this value into the given Serde serializer.