Model Evaluation Metrics

Classification Metrics

Multiple metrics evaluate classification models beyond accuracy.

Accuracy, Precision, Recall

accuracy_score(y_true, y_pred). precision_score(y_true, y_pred). recall_score(y_true, y_pred).

Precision: of predicted positives, how many are correct. Recall: of actual positives, how many were predicted.

F1 Score and AUC

F1 combines precision and recall: f1_score(y_true, y_pred). The harmonic mean balances both.

AUC (ROC AUC): roc_auc_score(y_true, y_proba). Measures discrimination across thresholds.

Confusion Matrix

confusion_matrix(y_true, y_pred) shows prediction breakdown. Classification report provides comprehensive metrics.

Key Takeaways

1. Accuracy is misleading for imbalanced data
2. Precision, recall, F1 provide nuanced evaluation
3. AUC evaluates ranking independent of threshold