Metrics API¶

Evaluation metrics for classification and regression models.

The foodspec.metrics module provides comprehensive performance metrics for model evaluation, including accuracy, precision, recall, ROC/PR curves, and regression metrics.

Classification Metrics¶

compute_classification_metrics¶

Comprehensive classification metrics (accuracy, precision, recall, F1).

Compute core classification metrics.

Parameters¶

y_true : array-like True labels. y_pred : array-like Predicted labels. labels : sequence, optional Class label order for confusion matrix. average : str, optional Averaging for precision/recall/F1 ('macro', 'micro', 'weighted'), by default 'macro'. y_scores : array-like, optional Probabilities or decision scores (binary) for ROC/PR.

Returns¶

dict accuracy, precision, recall, specificity, f1, balanced_accuracy, confusion_matrix, per_class metrics, optional roc/pr curves.

compute_roc_curve¶

Receiver Operating Characteristic curve data.

Convenience wrapper around sklearn roc_curve for binary tasks.

compute_pr_curve¶

Precision-Recall curve data.

Convenience wrapper around sklearn precision_recall_curve for binary tasks.

Regression Metrics¶

compute_regression_metrics¶

Comprehensive regression metrics (R², RMSE, MAE).

Compute regression metrics: RMSE, MAE, R2, MAPE, residuals.