Struct liblinear::util::TrainingInput

pub struct TrainingInput { /* fields omitted */ }

Input data for LibLinearModel and LibLinearCrossValidator.

Methods

impl TrainingInput

pub fn len_data(&self) -> usize

Number of training instances.

pub fn len_features(&self) -> usize

Dimensionality of the feature vector.

pub fn get(&self, index: usize) -> Option<&TrainingInstance>

Returns a reference to the training instance at the given index.

pub fn from_libsvm_file(path: &str) -> Result<TrainingInput, TrainingInputError>

Create a new instance from a LIBSVM training data file.

Each line in the data file represents a training instance and has the following format:

<target_value> <feature_index>:<feature_value> <feature_index>:<feature_value>...

Feature indices start from 1 and increase monotonically. However, they do not need to be continuous.

pub fn from_dense_features(
    labels: Vec<f64>,
    features: Vec<Vec<f64>>
) -> Result<TrainingInput, TrainingInputError>

Create a new instance from a dense vector of features and their corresponding target values.

pub fn from_sparse_features(
    labels: Vec<f64>,
    features: Vec<Vec<(u32, f64)>>
) -> Result<TrainingInput, TrainingInputError>

Create a new instance from a sparse vector of features and their corresponding target values.

The feature vector must be a list of tuples that encode a feature index and its corresponding feature value.

Trait Implementations

impl Default for TrainingInput

fn default() -> TrainingInput

Returns the "default value" for a type.

impl Clone for TrainingInput

fn clone(&self) -> TrainingInput

Returns a copy of the value.

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

Performs copy-assignment from source.