Struct juice::solver::regression_evaluator::RegressionEvaluator

pub struct RegressionEvaluator { /* private fields */ }

Sampled Evaluator for Regression Problems

Produces measure of accuracy for regression problems up to Capacity elements in a First-In First-Out stack.

Implementations

impl RegressionEvaluator

pub fn new(evaluation_metric: Option<String>) -> RegressionEvaluator

Create an evaluator for Regression Problems

Arguments

• evaluation_metric - Regression metric to use for evaluation - i.e. ‘mse’

pub fn add_sample(&mut self, prediction: f32, target: f32)

Add a sample by providing the expected target value and the prediction.

pub fn add_samples(&mut self, predictions: &[f32], targets: &[f32])

Add a batch of samples.

See add_sample.

pub fn get_predictions(&self, network_out: &mut SharedTensor<f32>) -> Vec<f32>

Get the predicted value from the output of a network.

pub fn set_capacity(&mut self, capacity: Option<usize>)

Set the capacity of the Regression Evaluator

pub fn samples(&self) -> &VecDeque<Sample>

Return all collected samples.

pub fn accuracy(&self) -> impl RegressionLoss

Return the accuracy of the collected predictions.

Trait Implementations

impl Debug for RegressionEvaluator

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

Formats the value using the given formatter.