ROC Curve & AUC Calculator

Overview

The ROC Curve & AUC Calculator provides comprehensive analysis of binary classification model performance. Plot Receiver Operating Characteristic (ROC) curves, calculate the Area Under the Curve (AUC), and explore the tradeoff between true positive rate and false positive rate at different classification thresholds. View confusion matrices, sensitivity, specificity, and predictive values at any threshold. This tool is essential for understanding and optimizing binary classifiers in medicine, fraud detection, and countless other applications.

Open in new tab

Tips

  • Use the threshold slider to see how classification metrics change across the full range
  • Perfect classification (AUC = 1.0) has the ROC curve hugging the top-left corner
  • Random guessing (AUC = 0.5) produces a diagonal line from bottom-left to top-right
  • The confusion matrix updates in real-time as you adjust the threshold
  • Watch how sensitivity (TPR) and specificity (TNR) trade off - increasing one often decreases the other
  • The optimal threshold depends on your use case - medical screening prefers high sensitivity, spam detection may prefer high specificity
  • Compare multiple ROC curves to evaluate different models or features
  • AUC gives a single number summary - higher is better, with 1.0 being perfect
  • PPV (precision) and NPV tell you the probability that positive/negative predictions are correct