Module classification 

Classification metrics module

This module provides functions for evaluating classification models, including accuracy, precision, recall, F1 score, ROC AUC, and advanced metrics.

Basic Metrics

Basic classification metrics include accuracy, precision, recall, and F1 score.

Advanced Metrics

Advanced metrics include Matthews Correlation Coefficient, balanced accuracy, Cohen’s kappa, Brier score, Jaccard similarity, and Hamming loss.

use scirs2_core::ndarray::array;
use scirs2_metrics::classification::advanced::{matthews_corrcoef, balanced_accuracy_score};

let y_true = array![0, 1, 2, 0, 1, 2];
let y_pred = array![0, 2, 1, 0, 0, 2];

let mcc = matthews_corrcoef(&y_true, &y_pred).unwrap();
let bal_acc = balanced_accuracy_score(&y_true, &y_pred).unwrap();

One-vs-One Metrics

One-vs-One metrics are useful for evaluating multi-class classification problems by considering each pair of classes separately.

use scirs2_core::ndarray::array;
use scirs2_metrics::classification::one_vs_one::{one_vs_one_accuracy, one_vs_one_f1_score};

let y_true = array![0, 1, 2, 0, 1, 2];
let y_pred = array![0, 2, 1, 0, 0, 2];

let ovo_acc = one_vs_one_accuracy(&y_true, &y_pred).unwrap();
let f1_scores = one_vs_one_f1_score(&y_true, &y_pred).unwrap();

Modules

advanced

Advanced classification metrics

curves

Curve functions for classification metrics

one_vs_one

One-vs-One Classification Metrics

threshold

Threshold optimization module

threshold_analyzer

Threshold analysis for binary classification

Functions

accuracy_score

Calculates accuracy score, the fraction of correctly classified samples

binary_log_loss

Calculate binary log loss, also known as binary cross-entropy

classification_report

Generates a text report showing the main classification metrics

confusion_matrix

Calculates a confusion matrix to evaluate the accuracy of a classification

f1_score

Calculates the F1 score for binary classification

fbeta_score

Calculates the F-beta score for binary classification

gain_chart

Computes the gain chart values for binary classification

lift_chart

Computes the lift chart values for binary classification

precision_score

Calculates the precision score for binary classification

recall_score

Calculates the recall score for binary classification

roc_auc_score

Compute Area Under the Receiver Operating Characteristic Curve (ROC AUC)