Interactive Confusion Matrix & Metrics Playground

1D Predicted Probability Distribution

Predicted Class

Positive (+)

Negative (-)

Total

Actual Positive

0 0.0% True Positive (TP)

0 0.0% False Negative (FN)

Actual Negative

0 0.0% False Positive (FP)

0 0.0% True Negative (TN)

Total

1. Add Custom Data Point (Click on Canvas above)

🟢 Actual Positive (Sick) 🔵 Actual Negative (Healthy)

2. Classification Threshold (τ)

Decision Cutoff Line0.50

Points to the right of this line are predicted as Positive. Points to the left are predicted as Negative.

3. Select Target Focus Class for Metrics Calculations

Positive Class (+) Negative Class (-)

Derived Evaluation Metrics Calculations

Accuracy

0.0%

Precision

0.0%

Sensitivity (Recall)

0.0%

Specificity

0.0%

🔬 Classroom Tradeoff Experiment: Switch the focus class back and forth. Notice how changing the perspective updates which outcomes constitute "Precision" or "Sensitivity", while global system **Accuracy** stays perfectly uniform!