Concept - Confusion Matrix
Definition (s.274): table comparing the true label with the predicted label for a binary classifier.
| Predicted Positive | Predicted Negative | |
|---|---|---|
| Positive Label | TP (True Positive) | FN (False Negative) |
| Negative Label | FP (False Positive) | TN (True Negative) |
- Accuracy = correct predictions / total samples (s.274).
- All other metrics are read off this table: Concept - Precision and Recall, Concept - F1 Score, TPR/FPR for Concept - ROC Curve and AUC.
- Example (H6751 auto-grade, s.276): TP=27, FN=4, FP=1, TN=18 -> accuracy 0.9.
- Accuracy is misleading on imbalanced data (s.280).
Related
Concept - Precision and Recall | Concept - F1 Score | Concept - ROC Curve and AUC
Source chunk: 03 Text Classification - Chunk 10 (s.271-300)