Struct LogisticRegression 

pub struct LogisticRegression<XT: RealNumber, YT: WholeNumber> { /* private fields */ }

Logistic regression model for binary classification.

This struct represents a logistic regression model for binary classification. It uses the sigmoid function to map the input features to a probability between 0 and 1, and makes predictions based on a threshold of 0.5.

Type Parameters

• XT: The type of the input features.
• YT: The type of the target labels.

Fields

• weights: The weights of the logistic regression model, with the first being the bias weight.
• _marker: A marker field to indicate the target label type.

Examples

use rusty_ai::regression::logistic::LogisticRegression;
use rusty_ai::data::dataset::Dataset;
use nalgebra::{DMatrix, DVector};

// Create a new logistic regression model
let mut model: LogisticRegression<f64, u8> = LogisticRegression::new();

// Fit the model to a dataset
let x = DMatrix::from_row_slice(3, 2, &[1.0, 2.0, 3.0, 4.0, 5.0, 6.0]);
let y = DVector::from_vec(vec![0, 1, 0]);
let dataset = Dataset::new(x, y);
let lr = 0.01;
let max_steps = 1000;
let epsilon = Some(0.001);
let progress = Some(100);
let result = model.fit(&dataset, lr, max_steps, epsilon, progress);

// Make predictions using the trained model
let x_pred = DMatrix::from_row_slice(2, 2, &[1.0, 2.0, 3.0, 4.0]);
let predictions = model.predict(&x_pred);

Implementations

impl<XT: RealNumber, YT: WholeNumber> LogisticRegression<XT, YT>

pub fn new() -> Self

Creates a new instance of LogisticRegression with default values.

Returns

A new LogisticRegression instance.

pub fn with_params( dimension: Option<usize>, weights: Option<DVector<XT>>, ) -> Result<Self, Box<dyn Error>>

Creates a new instance of LogisticRegression with custom parameters.

Parameters

• dimension: The dimension of the input features. If None, it will be inferred from the starting weights.
• weights: The starting weights for the logistic regression model. If None, default weights will be used.

Returns

A new LogisticRegression instance.

Errors

An error is returned if the dimension and weights are incompatible.

pub fn predict( &self, x_pred: &DMatrix<XT>, ) -> Result<DVector<YT>, Box<dyn Error>>

Predicts the target labels for the given input features.

Parameters

• x_pred: The input features to make predictions for.

Returns

A Result containing the predicted target labels if successful, or an error message if an error occurs during prediction.

pub fn fit( &mut self, dataset: &Dataset<XT, YT>, lr: XT, max_steps: usize, epsilon: Option<XT>, progress: Option<usize>, ) -> Result<String, Box<dyn Error>>

Fits the logistic regression model to a dataset.

Parameters

• dataset: The dataset to fit the model to.
• lr: The learning rate for gradient descent.
• max_steps: The maximum number of steps for gradient descent.
• epsilon: The convergence threshold for gradient descent. If None, a default value is used.
• progress: The number of steps to display progress information. If None, no progress is displayed.

Returns

A string indicating the result of the training process.

Errors

An error is returned if the progress steps value is 0.

pub fn weights(&self) -> &DVector<XT>

pub fn cross_entropy( &self, x: &DMatrix<XT>, y: &DVector<YT>, testing: bool, ) -> Result<XT, Box<dyn Error>>

Trait Implementations

impl<XT: RealNumber, YT: WholeNumber> ClassificationMetrics<YT> for LogisticRegression<XT, YT>

fn confusion_matrix( &self, y_true: &DVector<T>, y_pred: &DVector<T>, ) -> Result<DMatrix<usize>, Box<dyn Error>>

Computes the confusion matrix based on the true labels and predicted labels.

fn accuracy( &self, y_true: &DVector<T>, y_pred: &DVector<T>, ) -> Result<f64, Box<dyn Error>>

Computes the accuracy based on the true labels and predicted labels.

fn precision( &self, y_true: &DVector<T>, y_pred: &DVector<T>, ) -> Result<f64, Box<dyn Error>>

Computes the precision based on the true labels and predicted labels.

fn recall( &self, y_true: &DVector<T>, y_pred: &DVector<T>, ) -> Result<f64, Box<dyn Error>>

Computes the recall based on the true labels and predicted labels.

fn f1_score( &self, y_true: &DVector<T>, y_pred: &DVector<T>, ) -> Result<f64, Box<dyn Error>>

Computes the F1 score based on the true labels and predicted labels.

impl<XT: Clone + RealNumber, YT: Clone + WholeNumber> Clone for LogisticRegression<XT, YT>

fn clone(&self) -> LogisticRegression<XT, YT>

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl<XT: Debug + RealNumber, YT: Debug + WholeNumber> Debug for LogisticRegression<XT, YT>

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

Formats the value using the given formatter.

impl<XT: RealNumber, YT: WholeNumber> Default for LogisticRegression<XT, YT>

fn default() -> Self

Creates a new instance of LogisticRegression with default values.

Returns

A new LogisticRegression instance.